Method and apparatus for performing computer vision-based tasks

Abstract

A method for performing a computer vision-based task is proposed. The method comprises receiving an input image, and processing the input image at different resolutions to operate on image data of the input image at different scales by applying a dynamic scale pyramid that provides a variable number of scales representing the input image at different levels of resolution.

Description

[0001] METHOD AND APPARATUS FOR PERFORMING COMPUTER VISION-BASED TASKS

[0002] FIELD OF INVENTION

[0003] The present technology relates to a method, a neural network, and an apparatus for performing computer vision-based tasks, in particular, to a method, a neural network, and an apparatus providing an improved performance of the computer vision-based tasks.

[0004] BACKGROUND

[0005] Computer vision is a subfield of deep learning and artificial intelligence (Al). It refers to the processing, analysis, and interpretation of image data to enable machines such as computers to process complex real- world visual data, e.g., to correctly identify an object or person in a digital image and, if necessary, take appropriate action. Computer vision-based tasks may comprise image classification, object detection and localization, semantic segmentation, instance segmentation, pose estimation, image generation and synthesis, and pattern recognition.

[0006] In this context, the image data to be processed may take many forms, such as video sequences, views from multiple cameras, multi-dimensional data from a 3D scanner, 3D point clouds from LiDaR sensors, or medical scanning devices. It may comprise scenes in the world that include objects of many sizes, and these objects may contain features of many sizes. Moreover, objects can be at various distances from a viewer. In view of this, a corresponding method for performing a computer vision-based task may need to consider analyzing the image data at various levels of resolutions or scales. A common tool for performing a multiresolution analysis is related to a pyramid representation of the image data providing a collection of the same image with different resolutions. More specifically, an image pyramid is a data structure supporting a scaled convolution through reduced image representation. It consists of a sequence of copies of an original image in which both sample density and resolution are decreased in regular steps. This technique has been widely used for object detection methods. For example, feature pyramids built upon image pyramids form the basis of a standard solution approach when applied in neural networks, e.g., such as convolutional networks (ConvNets).

[0007] However, these common pyramid-based methods have some drawbacks. For example, featurizing each level of an image pyramid increases inference time of a neural network making this approach impractical for computer vision-based tasks such as object recognition. Moreover, training networks end-to-end on an image pyramid implies a high amount of memory. Moreover, in common applications implementing a pyramid in a neural network, the maximum number of scales is determined by the input resolution of an input image. Therefore, if multiple resolutions are to be supported, multiple networks with different resolution datasets need to be trained.

[0008] In addition, in general, the number of scales is pre-determined and fixed resulting in a higher computational cost. It is therefore desirable to improve a method for performing computer vision-based tasks to provide an optimized analysis of image data in terms of performance and / or computational efficiency.

[0009] SUMMARY OF INVENTION

[0010] To this end, a method for performing a computer vision-based task is provided that comprises receiving an input image, and processing the input image at different resolutions to operate on image data of the input image at different scales by applying a dynamic scale pyramid that provides a variable number of scales representing the input image at different levels of resolution.

[0011] Furthermore, a neural network is provided and configured to apply the method described above.

[0012] In addition, an apparatus comprising an image sensor, an event-based vision sensor, and a processing unit configured to execute the method is provided.

[0013] The use of a dynamic scale pyramid with a variable number of scales allows providing scales optimized for either performance or computational efficiency. More specifically, the number of scales can be increased or decreased to either improve task performance or reduce computational cost. Furthermore, by employing dynamic scaling, it is possible to train networks on a small resolution input and without re-training, enable it to run inference on a higher resolution input. This is beneficial for training times / resources or where high- resolution datasets are not available.

[0014] BRIEF DESCRIPTION OF DRAWINGS

[0015] Fig. 1 is a schematic diagram of a sensor device.

[0016] Fig. 2 is a schematic block diagram of a sensor section.

[0017] Fig. 3 is a schematic block diagram of a pixel array section.

[0018] Fig. 4 is a schematic circuit diagram of a pixel block.

[0019] Fig. 5 is a schematic block diagram illustrating of an event detecting section.

[0020] Fig. 6 is a schematic circuit diagram of a current-voltage converting section.

[0021] Fig. 7 is a schematic circuit diagram of a subtraction section and a quantization section.

[0022] Fig. 8 is a schematic timing chart of an example operation of the sensor section.

[0023] Fig. 9 is a schematic diagram of a frame data generation method based on event data. Fig. 10 is a schematic block diagram of another quantization section.

[0024] Fig. 11 is a schematic diagram of another event detecting section.

[0025] Fig. 12 is a schematic block diagram of another pixel array section.

[0026] Fig. 13 is a schematic circuit diagram of another pixel block.

[0027] Fig. 14 is a schematic block diagram of a scan-type imaging device.

[0028] Fig. 15 is a schematic diagram of an exemplary image processing applying a pyramid-based approach.

[0029] Fig. 16 is a schematic diagram of an exemplary image processing applying a further pyramid-based approach.

[0030] Fig. 17 illustrates schematically a process flow of a method for performing a computer vision-based task based on a dynamic scale pyramid.

[0031] Fig. 18 is a schematic diagram of an exemplary input image at different resolutions to be processed applying a fixed number of scales.

[0032] Fig. 19A is a schematic block diagram of an exemplary neural network.

[0033] Fig. 19B is a schematic block diagram of the exemplary neural network.

[0034] Fig. 20 is a schematic block diagram for performing a computer vision-based task applying an exemplary dynamic scale pyramid-based method.

[0035] Fig. 21 is a schematic block diagram for performing a computer vision-based task applying a further exemplary dynamic scale pyramid-based method.

[0036] Fig. 22 is a schematic block diagram for performing a computer vision-based task applying a further exemplary dynamic scale pyramid-based method.

[0037] Fig. 23 is a schematic block diagram of an exemplary apparatus.

[0038] Fig. 24 is a schematic block diagram of a vehicle control system.

[0039] Fig. 25 is a diagram of assistance in explaining an example of installation positions of an outside-vehicle information detecting section and an imaging section.

[0040] DETAILED DESCRIPTION The present disclosure is directed to mitigating problems regarding efficiency and computational costs of computer vision-based tasks. In particular, the problem is addressed how to improve task performance and / or reduce the computational cost. The solution to this problem comprises the appliance of a dynamical scale pyramid that provides a variable number of scales representing an input image at different levels of resolution. According to a first aspect, the scales may be increased or decreased based on the resolution of the input image. In addition, according to a second aspect, the scales may be dynamically adapted based on information related to the input image such as event-based data. The present disclosure is thus based on the operation of a conventional image sensor and a conventional event based / dynamic vision sensor (EVS / DVS).

[0041] Thus, at first a possible implementation of an EVS / DVS will be described. This is of course purely exemplary. It is to be understood that EVSs / DVSs could also be implemented differently.

[0042] Fig. 1 is a diagram illustrating a configuration example of a sensor device 10, which is in the example of Fig. 1 constituted by a sensor chip.

[0043] The sensor device 10 is a single-chip semiconductor chip and includes a sensor die (substrate) 11, which serves as a plurality of dies (substrates), and a logic die 12 that are stacked. Note that, the sensor device 10 can also include only a single die or three or more stacked dies.

[0044] In the sensor device 10 of Fig. 1, the sensor die 11 includes (a circuit serving as) a sensor section 21, and the logic die 12 includes a logic section 22. Note that, the sensor section 21 can be partly formed on the logic die 12. Further, the logic section 22 can be partly formed on the sensor die 11.

[0045] The sensor section 21 includes pixels configured to perform photoelectric conversion on incident light to generate electrical signals, and generates event data indicating the occurrence of events that are changes in the electrical signal of the pixels. The sensor section 21 supplies the event data to the logic section 22. That is, the sensor section 21 performs imaging of performing, in the pixels, photoelectric conversion on incident light to generate electrical signals, similarly to a synchronous image sensor, for example. The sensor section 21, however, generates event data indicating the occurrence of events that are changes in the electrical signal of the pixels instead of generating image data in a frame format (frame data). The sensor section 21 outputs, to the logic section 22, the event data obtained by the imaging.

[0046] Here, the synchronous image sensor is an image sensor configured to perform imaging in synchronization with a vertical synchronization signal and output frame data that is image data in a frame format. The sensor section 21 can be regarded as asynchronous (an asynchronous image sensor) in contrast to the synchronous image sensor since the sensor section 21 does not operate in synchronization with a vertical synchronization signal when outputting event data. Event detection may also be performed synchronously using a row scanner that determines which pixels generated an event during the past “frame” and assigns these events the same timestamp. Scan-type event detection is useful for moderate to high activity scenes since it incurs less readout overhead than arbiter-type, i.e. asynchronous, event detection. Note that, the sensor section 21 can generate and output, other than event data, frame data, similarly to the synchronous image sensor. In addition, the sensor section 21 can output, together with event data, electrical signals of pixels in which events have occurred, as pixel signals that are pixel values of the pixels in frame data.

[0047] The logic section 22 controls the sensor section 21 as needed. Further, the logic section 22 performs various types of data processing, such as data processing of generating frame data on the basis of event data from the sensor section 21 and image processing on frame data from the sensor section 21 or frame data generated on the basis of the event data from the sensor section 21, and outputs data processing results obtained by performing the various types of data processing on the event data and the frame data.

[0048] Fig. 2 is a block diagram illustrating a configuration example of the sensor section 21 of Fig. 1.

[0049] The sensor section 21 includes a pixel array section 31, a driving section 32, an arbiter 33, an AD (Analog to Digital) conversion section 34, and an output section 35.

[0050] The pixel array section 31 includes a plurality of pixels 51 (Fig. 3) arrayed in a two-dimensional lattice pattern. The pixel array section 31 detects, in a case where a change larger than a predetermined threshold (including a change equal to or larger than the threshold as needed) has occurred in (a voltage corresponding to) a photocurrent that is an electrical signal generated by photoelectric conversion in the pixel 51, the change in the photocurrent as an event. In a case of detecting an event, the pixel array section 31 outputs, to the arbiter 33, a request for requesting the output of event data indicating the occurrence of the event. Then, in a case of receiving a response indicating event data output permission from the arbiter 33, the pixel array section 31 outputs the event data to the driving section 32 and the output section 35. In addition, the pixel array section 31 outputs an electrical signal of the pixel 51 in which the event has been detected to the AD conversion section 34, as a pixel signal.

[0051] The driving section 32 supplies control signals to the pixel array section 31 to drive the pixel array section 31 . For example, the driving section 32 drives the pixel 51 regarding which the pixel array section 31 has output event data, so that the pixel 51 in question supplies (outputs) a pixel signal to the AD conversion section 34.

[0052] The arbiter 33 arbitrates the requests for requesting the output of event data from the pixel array section 31, and returns responses indicating event data output permission or prohibition to the pixel array section 31.

[0053] The AD conversion section 34 includes, for example, a single-slope ADC (AD converter) (not illustrated) in each column of pixel blocks 41 (Fig. 3) described later, for example. The AD conversion section 34 performs, with the ADC in each column, AD conversion on pixel signals of the pixels 51 of the pixel blocks 41 in the column, and supplies the resultant to the output section 35. Note that, the AD conversion section 34 can perform CDS (Correlated Double Sampling) together with pixel signal AD conversion. The output section 35 performs necessary processing on the pixel signals from the AD conversion section 34 and the event data from the pixel array section 31 and supplies the resultant to the logic section 22 (Fig. 1).

[0054] Here, a change in the photocurrent generated in the pixel 51 can be recognized as a change in the amount of light entering the pixel 51, so that it can also be said that an event is a change in light amount (a change in light amount larger than the threshold) in the pixel 51.

[0055] Event data indicating the occurrence of an event at least includes location information (coordinates or the like) indicating the location of a pixel block in which a change in light amount, which is the event, has occurred. Besides, the event data can also include the polarity (positive or negative) of the change in light amount.

[0056] With regard to the series of event data that is output from the pixel array section 31 at timings at which events have occurred, it can be said that, as long as the event data interval is the same as the event occurrence interval, the event data implicitly includes time point information indicating (relative) time points at which the events have occurred. However, for example, when the event data is stored in a memory and the event data interval is no longer the same as the event occurrence interval, the time point information implicitly included in the event data is lost. Thus, the output section 35 includes, in event data, time point information indicating (relative) time points at which events have occurred, such as timestamps, before the event data interval is changed from the event occurrence interval. The processing of including time point information in event data can be performed in any block other than the output section 35 as long as the processing is performed before time point information implicitly included in event data is lost.

[0057] Fig. 3 is a block diagram illustrating a configuration example of the pixel array section 31 of Fig. 2.

[0058] The pixel array section 31 includes the plurality of pixel blocks 41. The pixel block 41 includes the IxJ pixels 51 that are one or more pixels arrayed in I rows and J columns (I and J are integers), an event detecting section 52, and a pixel signal generating section 53. The one or more pixels 51 in the pixel block 41 share the event detecting section 52 and the pixel signal generating section 53. Further, in each column of the pixel blocks 41, a VSL (Vertical Signal Line) for connecting the pixel blocks 41 to the ADC of the AD conversion section 34 is wired.

[0059] The pixel 51 receives light incident from an object and performs photoelectric conversion to generate a photocurrent serving as an electrical signal. The pixel 51 supplies the photocurrent to the event detecting section 52 under the control of the driving section 32.

[0060] The event detecting section 52 detects, as an event, a change larger than the predetermined threshold in photocurrent from each of the pixels 51, under the control of the driving section 32. In a case of detecting an event, the event detecting section 52 supplies, to the arbiter 33 (Fig. 2), a request for requesting the output of event data indicating the occurrence of the event. Then, when receiving a response indicating event data output permission to the request from the arbiter 33, the event detecting section 52 outputs the event data to the driving section 32 and the output section 35.

[0061] The pixel signal generating section 53 generates, in the case where the event detecting section 52 has detected an event, a voltage corresponding to a photocurrent from the pixel 51 as a pixel signal, and supplies the voltage to the AD conversion section 34 through the VSL, under the control of the driving section 32.

[0062] Here, detecting a change larger than the predetermined threshold in photocurrent as an event can also be recognized as detecting, as an event, absence of change larger than the predetermined threshold in photocurrent. The pixel signal generating section 53 can generate a pixel signal in the case where absence of change larger than the predetermined threshold in photocurrent has been detected as an event as well as in the case where a change larger than the predetermined threshold in photocurrent has been detected as an event.

[0063] Fig. 4 is a circuit diagram illustrating a configuration example of the pixel block 41.

[0064] The pixel block 41 includes, as described with reference to Fig. 3, the pixels 51, the event detecting section 52, and the pixel signal generating section 53.

[0065] The pixel 51 includes a photoelectric conversion element 61 and transfer transistors 62 and 63.

[0066] The photoelectric conversion element 61 includes, for example, a PD (Photodiode). The photoelectric conversion element 61 receives incident light and performs photoelectric conversion to generate charges.

[0067] The transfer transistor 62 includes, for example, an N (Negative)-type MOS (Metal-Oxide-Semiconductor) FET (Field Effect Transistor). The transfer transistor 62 of the n-th pixel 51 of the lx J pixels 51 in the pixel block 41 is turned on or off in response to a control signal OFGn supplied from the driving section 32 (Fig. 2). When the transfer transistor 62 is turned on, charges generated in the photoelectric conversion element 61 are transferred (supplied) to the event detecting section 52, as a photocurrent.

[0068] The transfer transistor 63 includes, for example, an N-type MOSFET. The transfer transistor 63 of the n-th pixel 51 of the IxJ pixels 51 in the pixel block 41 is turned on or off in response to a control signal TRGn supplied from the driving section 32. When the transfer transistor 63 is turned on, charges generated in the photoelectric conversion element 61 are transferred to an FD 74 of the pixel signal generating section 53.

[0069] The IxJ pixels 51 in the pixel block 41 are connected to the event detecting section 52 of the pixel block 41 through nodes 60. Thus, photocurrents generated in (the photoelectric conversion elements 61 of) the pixels 51 are supplied to the event detecting section 52 through the nodes 60. As a result, the event detecting section 52 receives the sum of photocurrents from all the pixels 51 in the pixel block 41. Thus, the event detecting section 52 detects, as an event, a change in sum of photocurrents supplied from the IxJ pixels 51 in the pixel block 41. The pixel signal generating section 53 includes a reset transistor 71, an amplification transistor 72, a selection transistor 73, and the FD (Floating Diffusion) 74.

[0070] The reset transistor 71, the amplification transistor 72, and the selection transistor 73 include, for example, N-type MOSFETs.

[0071] The reset transistor 71 is turned on or off in response to a control signal RST supplied from the driving section 32 (Fig. 2). When the reset transistor 71 is turned on, the FD 74 is connected to a power supply VDD, and charges accumulated in the FD 74 are thus discharged to the power supply VDD. With this, the FD 74 is reset.

[0072] The amplification transistor 72 has a gate connected to the FD 74, a drain connected to the power supply VDD, and a source connected to the VSL through the selection transistor 73. The amplification transistor 72 is a source follower and outputs a voltage (electrical signal) corresponding to the voltage of the FD 74 supplied to the gate to the VSL through the selection transistor 73.

[0073] The selection transistor 73 is turned on or off in response to a control signal SEL supplied from the driving section 32. When the selection transistor 73 is turned on, a voltage corresponding to the voltage of the FD 74 from the amplification transistor 72 is output to the VSL.

[0074] The FD 74 accumulates charges transferred from the photoelectric conversion elements 61 of the pixels 51 through the transfer transistors 63, and converts the charges to voltages.

[0075] With regard to the pixels 51 and the pixel signal generating section 53, which are configured as described above, the driving section 32 turns on the transfer transistors 62 with control signals OFGn, so that the transfer transistors 62 supply, to the event detecting section 52, photocurrents based on charges generated in the photoelectric conversion elements 61 of the pixels 51. With this, the event detecting section 52 receives a current that is the sum of the photocurrents from all the pixels 51 in the pixel block 41, which might also be only a single pixel.

[0076] When the event detecting section 52 detects, as an event, a change in photocurrent (sum of photocurrents) in the pixel block 41, the driving section 32 turns off the transfer transistors 62 of all the pixels 51 in the pixel block 41, to thereby stop the supply of the photocurrents to the event detecting section 52. Then, the driving section 32 sequentially turns on, with the control signals TRGn, the transfer transistors 63 of the pixels 51 in the pixel block 41 in which the event has been detected, so that the transfer transistors 63 transfers charges generated in the photoelectric conversion elements 61 to the FD 74. The FD 74 accumulates the charges transferred from (the photoelectric conversion elements 61 of) the pixels 51. Voltages corresponding to the charges accumulated in the FD 74 are output to the VSL, as pixel signals of the pixels 51, through the amplification transistor 72 and the selection transistor 73. As described above, in the sensor section 21 (Fig. 2), only pixel signals of the pixels 51 in the pixel block 41 in which an event has been detected are sequentially output to the VSL. The pixel signals output to the VSL are supplied to the AD conversion section 34 to be subjected to AD conversion.

[0077] Here, in the pixels 51 in the pixel block 41, the transfer transistors 63 can be turned on not sequentially but simultaneously. In this case, the sum of pixel signals of all the pixels 51 in the pixel block 41 can be output.

[0078] In the pixel array section 31 of Fig. 3, the pixel block 41 includes one or more pixels 51, and the one or more pixels 51 share the event detecting section 52 and the pixel signal generating section 53. Thus, in the case where the pixel block 41 includes a plurality of pixels 51, the numbers of the event detecting sections 52 and the pixel signal generating sections 53 can be reduced as compared to a case where the event detecting section 52 and the pixel signal generating section 53 are provided for each of the pixels 51, with the result that the scale of the pixel array section 31 can be reduced.

[0079] Note that, in the case where the pixel block 41 includes a plurality of pixels 51, the event detecting section 52 can be provided for each of the pixels 51. In the case where the plurality of pixels 51 in the pixel block 41 share the event detecting section 52, events are detected in units of the pixel blocks 41. In the case where the event detecting section 52 is provided for each of the pixels 51, however, events can be detected in units of the pixels 51.

[0080] Yet, even in the case where the plurality of pixels 51 in the pixel block 41 share the single event detecting section 52, events can be detected in units of the pixels 51 when the transfer transistors 62 of the plurality of pixels 51 are temporarily turned on in a time-division manner.

[0081] Further, in a case where there is no need to output pixel signals, the pixel block 41 can be formed without the pixel signal generating section 53. In the case where the pixel block 41 is formed without the pixel signal generating section 53, the sensor section 21 can be formed without the AD conversion section 34 and the transfer transistors 63. In this case, the scale of the sensor section 21 can be reduced. The sensor will then output the address of the pixel (block) in which the event occurred, if necessary with a time stamp.

[0082] Fig. 5 is a block diagram illustrating a configuration example of the event detecting section 52 of Fig. 3.

[0083] The event detecting section 52 includes a current-voltage converting section 81, a buffer 82, a subtraction section 83, a quantization section 84, and a transfer section 85.

[0084] The current-voltage converting section 81 converts (a sum of) photocurrents from the pixels 51 to voltages corresponding to the logarithms of the photocurrents (hereinafter also referred to as a "photovoltage") and supplies the voltages to the buffer 82.

[0085] The buffer 82 buffers photovoltages from the current-voltage converting section 81 and supplies the resultant to the subtraction section 83. The subtraction section 83 calculates, at a timing instructed by a row driving signal that is a control signal from the driving section 32, a difference between the current photovoltage and a photovoltage at a timing slightly shifted from the current time, and supplies a difference signal corresponding to the difference to the quantization section 84.

[0086] The quantization section 84 quantizes difference signals from the subtraction section 83 to digital signals and supplies the quantized values of the difference signals to the transfer section 85 as event data.

[0087] The transfer section 85 transfers (outputs), on the basis of event data from the quantization section 84, the event data to the output section 35. That is, the transfer section 85 supplies a request for requesting the output of the event data to the arbiter 33. Then, when receiving a response indicating event data output permission to the request from the arbiter 33, the transfer section 85 outputs the event data to the output section 35.

[0088] Fig. 6 is a circuit diagram illustrating a configuration example of the current-voltage converting section 81 of Fig. 5.

[0089] The current-voltage converting section 81 includes transistors 91 to 93. As the transistors 91 and 93, for example, N-type MOSFETs can be employed. As the transistor 92, for example, a P-type MOSFET can be employed.

[0090] The transistor 91 has a source connected to the gate of the transistor 93, and a photocurrent is supplied from the pixel 51 to the connecting point between the source of the transistor 91 and the gate of the transistor 93. The transistor 91 has a drain connected to the power supply VDD and a gate connected to the drain of the transistor 93.

[0091] The transistor 92 has a source connected to the power supply VDD and a drain connected to the connecting point between the gate of the transistor 91 and the drain of the transistor 93. A predetermined bias voltage Vbias is applied to the gate of the transistor 92. With the bias voltage Vbias, the transistor 92 is turned on or off, and the operation of the current-voltage converting section 81 is turned on or off depending on whether the transistor 92 is turned on or off.

[0092] The source of the transistor 93 is grounded.

[0093] In the current-voltage converting section 81, the transistor 91 has the drain connected on the power supply VDD side and is thus a source follower. The source of the transistor 91, which is the source follower, is connected to the pixels 51 (Fig. 4), so that photocurrents based on charges generated in the photoelectric conversion elements 61 of the pixels 51 flow through the transistor 91 (from the drain to the source). The transistor 91 operates in a subthreshold region, and at the gate of the transistor 91, photovoltages corresponding to the logarithms of the photocurrents flowing through the transistor 91 are generated. As described above, in the current-voltage converting section 81, the transistor 91 converts photocurrents from the pixels 51 to photovoltages corresponding to the logarithms of the photocurrents. In the current-voltage converting section 81 , the transistor 91 has the gate connected to the connecting point between the drain of the transistor 92 and the drain of the transistor 93, and the photovoltages are output from the connecting point in question.

[0094] Fig. 7 is a circuit diagram illustrating configuration examples of the subtraction section 83 and the quantization section 84 of Fig. 5.

[0095] The subtraction section 83 includes a capacitor 101, an operational amplifier 102, a capacitor 103, and a switch 104. The quantization section 84 includes a comparator 111.

[0096] The capacitor 101 has one end connected to the output terminal of the buffer 82 (Fig. 5) and the other end connected to the input terminal (inverting input terminal) of the operational amplifier 102. Thus, photovoltages are input to the input terminal of the operational amplifier 102 through the capacitor 101.

[0097] The operational amplifier 102 has an output terminal connected to the non-inverting input terminal (+) of the comparator 111.

[0098] The capacitor 103 has one end connected to the input terminal of the operational amplifier 102 and the other end connected to the output terminal of the operational amplifier 102.

[0099] The switch 104 is connected to the capacitor 103 to switch the connections between the ends of the capacitor 103. The switch 104 is turned on or off in response to a row driving signal that is a control signal from the driving section 32, to thereby switch the connections between the ends of the capacitor 103.

[0100] A photovoltage on the buffer 82 (Fig. 5) side of the capacitor 101 when the switch 104 is on is denoted by Vinit, and the capacitance (electrostatic capacitance) of the capacitor 101 is denoted by Cl. The input terminal of the operational amplifier 102 serves as a virtual ground terminal, and a charge Qinit that is accumulated in the capacitor 101 in the case where the switch 104 is on is expressed by Expression (1).

[0101] Qinit = Cl x Vinit (1)

[0102] Further, in the case where the switch 104 is on, the connection between the ends of the capacitor 103 is cut (short-circuited), so that no charge is accumulated in the capacitor 103.

[0103] When a photovoltage on the buffer 82 (Fig. 5) side of the capacitor 101 in the case where the switch 104 has thereafter been turned off is denoted by Vafter, a charge Qafter that is accumulated in the capacitor 101 in the case where the switch 104 is off is expressed by Expression (2).

[0104] Qafter = C 1 x Vafter (2) When the capacitance of the capacitor 103 is denoted by C2 and the output voltage of the operational amplifier 102 is denoted by Vout, a charge Q2 that is accumulated in the capacitor 103 is expressed by Expression (3).

[0105] Q2 = -C2 x Vout (3)

[0106] Since the total amount of charges in the capacitors 101 and 103 does not change before and after the switch 104 is turned off, Expression (4) is established.

[0107] Qinit = Qafter + Q2 (4)

[0108] When Expression (1) to Expression (3) are substituted for Expression (4), Expression (5) is obtained.

[0109] Vout = -(C1 / C2) x (Vafter - Vinit) (5)

[0110] With Expression (5), the subtraction section 83 subtracts the photovoltage Vinit from the photovoltage Vafter, that is, calculates the difference signal (Vout) corresponding to a difference Vafter - Vinit between the photovoltages Vafter and Vinit. With Expression (5), the subtraction gain of the subtraction section 83 is C1 / C2. Since the maximum gain is normally desired, Cl is preferably set to a large value and C2 is preferably set to a small value. Meanwhile, when C2 is too small, kTC noise increases, resulting in a risk of deteriorated noise characteristics. Thus, the capacitance C2 can only be reduced in a range that achieves acceptable noise. Further, since the pixel blocks 41 each have installed therein the event detecting section 52 including the subtraction section 83, the capacitances Cl and C2 have space constraints. In consideration of these matters, the values of the capacitances Cl and C2 are determined.

[0111] The comparator 111 compares a difference signal from the subtraction section 83 with a predetermined threshold (voltage) Vth (>0) applied to the inverting input terminal (-), thereby quantizing the difference signal. The comparator 111 outputs the quantized value obtained by the quantization to the transfer section 85 as event data.

[0112] For example, in a case where a difference signal is larger than the threshold Vth, the comparator 111 outputs an H (High) level indicating 1, as event data indicating the occurrence of an event. In a case where a difference signal is not larger than the threshold Vth, the comparator 111 outputs an L (Low) level indicating 0, as event data indicating that no event has occurred.

[0113] The transfer section 85 supplies a request to the arbiter 33 in a case where it is confirmed on the basis of event data from the quantization section 84 that a change in light amount that is an event has occurred, that is, in the case where the difference signal (Vout) is larger than the threshold Vth. When receiving a response indicating event data output permission, the transfer section 85 outputs the event data indicating the occurrence of the event (for example, H level) to the output section 35. The output section 35 includes, in event data from the transfer section 85, location / address information regarding (the pixel block 41 including) the pixel 51 in which an event indicated by the event data has occurred and time point information indicating a time point at which the event has occurred, and further, as needed, the polarity of a change in light amount that is the event, i.e. whether the intensity did increase or decrease. The output section 35 outputs the event data.

[0114] As the data format of event data including location information regarding the pixel 51 in which an event has occurred, time point information indicating a time point at which the event has occurred, and the polarity of a change in light amount that is the event, for example, the data format called "AER (Address Event Representation)" can be employed.

[0115] Note that, a gain A of the entire event detecting section 52 is expressed by the following expression where the gain of the current-voltage converting section 81 is denoted by CGlog and the gain of the buffer 82 is 1.

[0116] A = CGlogCl / C2 (Eiphoto n) (6)

[0117] Here, iphoto n denotes a photocurrent of the n-th pixel 51 of the I* J pixels 51 in the pixel block 41. In Expression (6), E denotes the summation of n that takes integers ranging from 1 to IxJ.

[0118] Note that, the pixel 51 can receive any light as incident light with an optical fdter through which predetermined light passes, such as a color filter. For example, in a case where the pixel 51 receives visible light as incident light, event data indicates the occurrence of changes in pixel value in images including visible objects. Further, for example, in a case where the pixel 51 receives, as incident light, infrared light, millimeter waves, or the like for ranging, event data indicates the occurrence of changes in distances to objects. In addition, for example, in a case where the pixel 51 receives infrared light for temperature measurement, as incident light, event data indicates the occurrence of changes in temperature of objects. In the present embodiment, the pixel 51 is assumed to receive visible light as incident light.

[0119] Fig. 8 is a timing chart illustrating an example of the operation of the sensor section 21 of Fig. 2.

[0120] At Timing TO, the driving section 32 changes all the control signals OFGn from the L level to the H level, thereby turning on the transfer transistors 62 of all the pixels 51 in the pixel block 41. With this, the sum of photocurrents from all the pixels 51 in the pixel block 41 is supplied to the event detecting section 52. Here, the control signals TRGn are all at the L level, and hence the transfer transistors 63 of all the pixels 51 are off.

[0121] For example, at Timing Tl, when detecting an event, the event detecting section 52 outputs event data at the H level in response to the detection of the event.

[0122] At Timing T2, the driving section 32 sets all the control signals OFGn to the L level on the basis of the event data at the H level, to stop the supply of the photocurrents from the pixels 51 to the event detecting section 52. Further, the driving section 32 sets the control signal SEL to the H level, and sets the control signal RST to the H level over a certain period of time, to control the FD 74 to discharge the charges to the power supply VDD, thereby resetting the FD 74. The pixel signal generating section 53 outputs, as a reset level, a pixel signal corresponding to the voltage of the FD 74 when the FD 74 has been reset, and the AD conversion section 34 performs AD conversion on the reset level.

[0123] At Timing T3 after the reset level AD conversion, the driving section 32 sets a control signal TRG1 to the H level over a certain period to control the first pixel 51 in the pixel block 41 in which the event has been detected to transfer, to the FD 74, charges generated by photoelectric conversion in (the photoelectric conversion element 61 of) the first pixel 51. The pixel signal generating section 53 outputs, as a signal level, a pixel signal corresponding to the voltage of the FD 74 to which the charges have been transferred from the pixel 51, and the AD conversion section 34 performs AD conversion on the signal level.

[0124] The AD conversion section 34 outputs, to the output section 35, a difference between the signal level and the reset level obtained after the AD conversion, as a pixel signal serving as a pixel value of the image (frame data).

[0125] Here, the processing of obtaining a difference between a signal level and a reset level as a pixel signal serving as a pixel value of an image is called "CDS." CDS can be performed after the AD conversion of a signal level and a reset level, or can be simultaneously performed with the AD conversion of a signal level and a reset level in a case where the AD conversion section 34 performs single-slope AD conversion. In the latter case, AD conversion is performed on the signal level by using the AD conversion result of the reset level as an initial value.

[0126] At Timing T4 after the AD conversion of the pixel signal of the first pixel 51 in the pixel block 41, the driving section 32 sets a control signal TRG2 to the H level over a certain period of time to control the second pixel 51 in the pixel block 41 in which the event has been detected to output a pixel signal.

[0127] In the sensor section 21, similar processing is executed thereafter, so that pixel signals of the pixels 51 in the pixel block 41 in which the event has been detected are sequentially output.

[0128] When the pixel signals of all the pixels 51 in the pixel block 41 are output, the driving section 32 sets all the control signals OFGn to the H level to turn on the transfer transistors 62 of all the pixels 51 in the pixel block 41.

[0129] Fig. 9 is a diagram illustrating an example of a frame data generation method based on event data.

[0130] The logic section 22 sets a frame interval and a frame width on the basis of an externally input command, for example. Here, the frame interval represents the interval of frames of frame data that is generated on the basis of event data. The frame width represents the time width of event data that is used for generating frame data on a single frame. A frame interval and a frame width that are set by the logic section 22 are also referred to as a "set frame interval" and a "set frame width," respectively. The logic section 22 generates, on the basis of the set frame interval, the set frame width, and event data from the sensor section 21, frame data that is image data in a frame format, to thereby convert the event data to the frame data.

[0131] That is, the logic section 22 generates, in each set frame interval, frame data on the basis of event data in the set frame width from the beginning of the set frame interval.

[0132] Here, it is assumed that event data includes time point information ti indicating a time point at which an event has occurred (hereinafter also referred to as an "event time point") and coordinates (x, y) serving as location information regarding (the pixel block 41 including) the pixel 51 in which the event has occurred (hereinafter also referred to as an "event location").

[0133] In Fig. 9, in a three-dimensional space (time and space) with the x axis, the y axis, and the time axis t, points representing event data are plotted on the basis of the event time point t and the event location (coordinates) (x, y) included in the event data.

[0134] That is, when a location (x, y, t) on the three-dimensional space indicated by the event time point t and the event location (x, y) included in event data is regarded as the space-time location of an event, in Fig. 9, the points representing the event data are plotted on the space-time locations (x, y, t) of the events.

[0135] The logic section 22 starts to generate frame data on the basis of event data by using, as a generation start time point at which frame data generation starts, a predetermined time point, for example, a time point at which frame data generation is externally instructed or a time point at which the sensor device 10 is powered on.

[0136] Here, cuboids each having the set frame width in the direction of the time axis t in the set frame intervals, which appear from the generation start time point, are referred to as a "frame volume." The size of the frame volume in the x-axis direction or the y-axis direction is equal to the number of the pixel blocks 41 or the pixels 51 in the x-axis direction or the y-axis direction, for example.

[0137] The logic section 22 generates, in each set frame interval, frame data on a single frame on the basis of event data in the frame volume having the set frame width from the beginning of the set frame interval.

[0138] Frame data can be generated by, for example, setting white to a pixel (pixel value) in a frame at the event location (x, y) included in event data and setting a predetermined color such as gray to pixels at other locations in the frame.

[0139] Besides, in a case where event data includes the polarity of a change in light amount that is an event, frame data can be generated in consideration of the polarity included in the event data. For example, white can be set to pixels in the case a positive polarity, while black can be set to pixels in the case of a negative polarity. In addition, in the case where pixel signals of the pixels 51 are also output when event data is output as described with reference to Fig. 3 and Fig. 4, frame data can be generated on the basis of the event data by using the pixel signals of the pixels 51 . That is, frame data can be generated by setting, in a frame, a pixel at the event location (x, y) (in a block corresponding to the pixel block 41) included in event data to a pixel signal of the pixel 51 at the location (x, y) and setting a predetermined color such as gray to pixels at other locations.

[0140] Note that, in the frame volume, there are a plurality of pieces of event data that are different in the event time point t but the same in the event location (x, y) in some cases. In this case, for example, event data at the latest or oldest event time point t can be prioritized. Further, in the case where event data includes polarities, the polarities of a plurality of pieces of event data that are different in the event time point t but the same in the event location (x, y) can be added together, and a pixel value based on the added value obtained by the addition can be set to a pixel at the event location (x, y).

[0141] Here, in a case where the frame width and the frame interval are the same, the frame volumes are adjacent to each other without any gap. Further, in a case where the frame interval is larger than the frame width, the frame volumes are arranged with gaps. In a case where the frame width is larger than the frame interval, the frame volumes are arranged to be partly overlapped with each other.

[0142] Fig. 10 is a block diagram illustrating another configuration example of the quantization section 84 of Fig. 5.

[0143] Note that, in Fig. 10, parts corresponding to those in the case of Fig. 7 are denoted by the same reference signs, and the description thereof is omitted as appropriate below.

[0144] In Fig. 10, the quantization section 84 includes comparators 111 and 112 and an output section 113.

[0145] Thus, the quantization section 84 of Fig. 10 is similar to the case of Fig. 7 in including the comparator 111. However, the quantization section 84 of Fig. 10 is different from the case of Fig. 7 in newly including the comparator 112 and the output section 113.

[0146] The event detecting section 52 (Fig. 5) including the quantization section 84 of Fig. 10 detects, in addition to events, the polarities of changes in light amount that are events.

[0147] In the quantization section 84 of Fig. 10, the comparator 111 outputs, in the case where a difference signal is larger than the threshold Vth, the H level indicating 1, as event data indicating the occurrence of an event having the positive polarity. The comparator 111 outputs, in the case where a difference signal is not larger than the threshold Vth, the L level indicating 0, as event data indicating that no event having the positive polarity has occurred. Further, in the quantization section 84 of Fig. 10, a threshold Vth' (<Vth) is supplied to the non-inverting input terminal (+) of the comparator 112, and difference signals are supplied to the inverting input terminal (-) of the comparator 112 from the subtraction section 83. Here, for the sake of simple description, it is assumed that the threshold Vth' is equal to -Vth, for example, which needs however not to be the case.

[0148] The comparator 112 compares a difference signal from the subtraction section 83 with the threshold Vth' applied to the inverting input terminal (-), thereby quantizing the difference signal. The comparator 112 outputs, as event data, the quantized value obtained by the quantization.

[0149] For example, in a case where a difference signal is smaller than the threshold Vth' (the absolute value of the difference signal having a negative value is larger than the threshold Vth), the comparator 112 outputs the H level indicating 1, as event data indicating the occurrence of an event having the negative polarity. Further, in a case where a difference signal is not smaller than the threshold Vth' (the absolute value of the difference signal having a negative value is not larger than the threshold Vth), the comparator 112 outputs the L level indicating 0, as event data indicating that no event having the negative polarity has occurred.

[0150] The output section 113 outputs, on the basis of event data output from the comparators 111 and 112, event data indicating the occurrence of an event having the positive polarity, event data indicating the occurrence of an event having the negative polarity, or event data indicating that no event has occurred to the transfer section 85.

[0151] For example, the output section 113 outputs, in a case where event data from the comparator 111 is the H level indicating 1, +V volts indicating +1, as event data indicating the occurrence of an event having the positive polarity, to the transfer section 85. Further, the output section 113 outputs, in a case where event data from the comparator 112 is the H level indicating 1, -V volts indicating -1, as event data indicating the occurrence of an event having the negative polarity, to the transfer section 85. In addition, the output section 113 outputs, in a case where each event data from the comparators 111 and 112 is the L level indicating 0, 0 volts (GND level) indicating 0, as event data indicating that no event has occurred, to the transfer section 85.

[0152] The transfer section 85 supplies a request to the arbiter 33 in the case where it is confirmed on the basis of event data from the output section 113 of the quantization section 84 that a change in light amount that is an event having the positive polarity or the negative polarity has occurred. After receiving a response indicating event data output permission, the transfer section 85 outputs event data indicating the occurrence of the event having the positive polarity or the negative polarity (+V volts indicating 1 or -V volts indicating -1) to the output section 35.

[0153] Preferably, the quantization section 84 has a configuration as illustrated in Fig. 10.

[0154] Fig. 11 is a diagram illustrating another configuration example of the event detecting section 52. In Fig. 11, the event detecting section 52 includes a subtractor 430, a quantizer 440, a memory 451, and a controller 452. The subtractor 430 and the quantizer 440 correspond to the subtraction section 83 and the quantization section 84, respectively.

[0155] Note that, in Fig. 11, the event detecting section 52 further includes blocks corresponding to the currentvoltage converting section 81 and the buffer 82, but the illustrations of the blocks are omitted in Fig. 11.

[0156] The subtractor 430 includes a capacitor 431, an operational amplifier 432, a capacitor 433, and a switch 434. The capacitor 431, the operational amplifier 432, the capacitor 433, and the switch 434 correspond to the capacitor 101, the operational amplifier 102, the capacitor 103, and the switch 104, respectively.

[0157] The quantizer 440 includes a comparator 441. The comparator 441 corresponds to the comparator 111.

[0158] The comparator 441 compares a voltage signal (difference signal) from the subtractor 430 with the predetermined threshold voltage Vth applied to the inverting input terminal (-). The comparator 441 outputs a signal indicating the comparison result, as a detection signal (quantized value).

[0159] The voltage signal from the subtractor 430 may be input to the input terminal (-) of the comparator 441, and the predetermined threshold voltage Vth may be input to the input terminal (+) of the comparator 441.

[0160] The controller 452 supplies the predetermined threshold voltage Vth applied to the inverting input terminal (-) of the comparator 441. The threshold voltage Vth which is supplied may be changed in a time-division manner. For example, the controller 452 supplies a threshold voltage Vthl corresponding to ON events (for example, positive changes in photocurrent) and a threshold voltage Vth2 corresponding to OFF events (for example, negative changes in photocurrent) at different timings to allow the single comparator to detect a plurality of types of address events (events).

[0161] The memory 451 accumulates output from the comparator 441 on the basis of Sample signals supplied from the controller 452. The memory 451 may be a sampling circuit, such as a switch, plastic, or capacitor, or a digital memory circuit, such as a latch or flip-flop. For example, the memory 451 may hold, in a period in which the threshold voltage Vth2 corresponding to OFF events is supplied to the inverting input terminal (-) of the comparator 441, the result of comparison by the comparator 441 using the threshold voltage Vthl corresponding to ON events. Note that, the memory 451 may be omitted, may be provided inside the pixel (pixel block 41), or may be provided outside the pixel.

[0162] Fig. 12 is a block diagram illustrating another configuration example of the pixel array section 31 of Fig. 2.

[0163] Note that, in Fig. 12, parts corresponding to those in the case of Fig. 3 are denoted by the same reference signs, and the description thereof is omitted as appropriate below. In Fig. 12, the pixel array section 31 includes the plurality of pixel blocks 41. The pixel block 41 includes the lx J pixels 51 that are one or more pixels and the event detecting section 52.

[0164] Thus, the pixel array section 31 of Fig. 12 is similar to the case of Fig. 3 in that the pixel array section 31 includes the plurality of pixel blocks 41 and that the pixel block 41 includes one or more pixels 51 and the event detecting section 52. However, the pixel array section 31 of Fig. 12 is different from the case of Fig. 3 in that the pixel block 41 does not include the pixel signal generating section 53.

[0165] As described above, in the pixel array section 31 of Fig. 12, the pixel block 41 does not include the pixel signal generating section 53, so that the sensor section 21 (Fig. 2) can be formed without the AD conversion section 34.

[0166] Fig. 13 is a circuit diagram illustrating a configuration example of the pixel block 41 of Fig. 12.

[0167] As described with reference to Fig. 12, the pixel block 41 includes the pixels 51 and the event detecting section 52, but does not include the pixel signal generating section 53.

[0168] In this case, the pixel 51 can only include the photoelectric conversion element 61 without the transfer transistors 62 and 63.

[0169] Note that, in the case where the pixel 51 has the configuration illustrated in Fig. 13, the event detecting section 52 can output a voltage corresponding to a photocurrent from the pixel 51, as a pixel signal.

[0170] Above, the sensor device 10 was described to be an asynchronous imaging device configured to read out events by the asynchronous readout system. However, the event readout system is not limited to the asynchronous readout system and may be the synchronous readout system. An imaging device to which the synchronous readout system is applied is a scan type imaging device that is the same as a general imaging device configured to perform imaging at a predetermined frame rate.

[0171] Fig. 14 is a block diagram illustrating a configuration example of a scan type imaging device.

[0172] As illustrated in Fig. 14, an imaging device 510 includes a pixel array section 521, a driving section 522, a signal processing section 525, a read-out region selecting section 527, and a signal generating section 528.

[0173] The pixel array section 521 includes a plurality of pixels 530. The plurality of pixels 530 each output an output signal in response to a selection signal from the read-out region selecting section 527. The plurality of pixels 530 can each include an in-pixel quantizer as illustrated in Fig. 11, for example. The plurality of pixels 530 output output signals corresponding to the amounts of change in light intensity. The plurality of pixels 530 may be two-dimensionally disposed in a matrix as illustrated in Fig. 14.

[0174] The driving section 522 drives the plurality of pixels 530, so that the pixels 530 output pixel signals generated in the pixels 530 to the signal processing section 525 through an output line 514. Note that, the driving section 522 and the signal processing section 525 are circuit sections for acquiring grayscale information. Thus, in a case where only event information (event data) is acquired, the driving section 522 and the signal processing section 525 may be omitted.

[0175] The read-out region selecting section 527 selects some of the plurality of pixels 530 included in the pixel array section 521. For example, the read-out region selecting section 527 selects one or a plurality of rows included in the two-dimensional matrix structure corresponding to the pixel array section 521. The readout region selecting section 527 sequentially selects one or a plurality of rows on the basis of a cycle set in advance. Further, the read-out region selecting section 527 may determine a selection region on the basis of requests from the pixels 530 in the pixel array section 521.

[0176] The signal generating section 528 generates, on the basis of output signals of the pixels 530 selected by the read-out region selecting section 527, event signals corresponding to active pixels in which events have been detected of the selected pixels 530. The events mean an event that the intensity of light changes. The active pixels mean the pixel 530 in which the amount of change in light intensity corresponding to an output signal exceeds or falls below a threshold set in advance. For example, the signal generating section 528 compares output signals from the pixels 530 with a reference signal, and detects, as an active pixel, a pixel that outputs an output signal larger or smaller than the reference signal. The signal generating section 528 generates an event signal (event data) corresponding to the active pixel.

[0177] The signal generating section 528 can include, for example, a column selecting circuit configured to arbitrate signals input to the signal generating section 528. Further, the signal generating section 528 can output not only information regarding active pixels in which events have been detected, but also information regarding non-active pixels in which no event has been detected.

[0178] The signal generating section 528 outputs, through an output line 515, address information and timestamp information (for example, (X, Y, T)) regarding the active pixels in which the events have been detected. However, the data that is output from the signal generating section 528 may not only be the address information and the timestamp information, but also information in a frame format (for example, (0, 0, 1, o, -)).

[0179] In the following description reference will mainly be made to sensor devices of the EVS type as described above in order to ease the description and to cover an important application example. However, the principles explained below apply just as well to any event-based vision sensor that is capable to generate events based on the occurrence of temporal intensity changes. In the following it will be described how this basic concept of event-based vision sensors can be extended to color variations.

[0180] Fig. 15 shows the general principle of a pyramid-based approach for image processing. More specifically, in the upper part of Fig. 15, an image pyramid GSP is illustrated in a simplified manner, which represents an input image Io at different scales SCo, SCi, SC2, SC3 or levels of resolution. At each scale SCi, SC2, SC3, the input image Io is represented by a copy Ii, E, I3 of itself having a different, i.e., decreasing, resolution. This is achieved by smoothing the input image Io with an appropriate smoothing filter or kernel and then subsampling (down sampling) the smoothed image, usually by a factor of 2 along each coordinate direction x, y. The resulting image Ii, is then subjected to the same procedure, and the cycle is repeated multiple times. Each cycle of this process results in a smaller image E, E with increased smoothing, but with decreased spatial sampling density, that is, decreased image resolution (half of width and height). The resulting multi-scale representation looks like a pyramid as shown in the upper part of Fig. 15, with the original input image Io on the bottom and the smaller images Ii, E, E resulting from the cycles stacked one atop the other. The number of cycles and thus, the resulting number of copied images is not limited to the ones illustrated in the example of Fig. 15. More cycles may be performed.

[0181] The lower part of Fig. 15 shows a specific example for the input image Io and its sequence of copies Ii, E, I3. As can be seen, the resolution decreases with increasing scales SCo, SCi, SC2, SC3 starting from the original input image Io which has a high resolution Hres towards the image I3 which has the lowest resolution Lres.

[0182] The pyramid construction described above is equivalent to convolving the original input image Io with a set of Gaussian-like weighting functions. The convolution acts as a lowpass filter with a band limit reduced correspondingly by one octave with each scale SCi, SC2, SC3 or level. Therefore, the multi-scale representation is also referred to as a lowpass pyramid or a Gaussian pyramid GSP.

[0183] In addition to the lowpass pyramid or Gaussian pyramid GSP, a bandpass pyramid may be made by forming the difference between images Ii, E, I3 at adjacent scales SCi, SC2, SC3 in the pyramid GSP and performing image interpolation between adjacent scales SCi, SC2, SC3 of resolution, to enable computation of pixelwise differences. This is shown in a simplified manner in Fig. 16.

[0184] Fig. 16 is a schematic diagram of an exemplary image processing applying a further pyramid-based approach. The bandpass images shown on the right side of Fig. 16 may be obtained by subtracting the image I3, E, Ii corresponding to a respective scale SC3, SC2, SCi or level of the pyramid GSP from the image E, Ii, Io of the next-lower scale SC2, SCi, SCo in the pyramid GSP. Only the smallest scale SC3 is not a difference image to enable reconstruction of the input image Io. Because these scales SC3, SC2, SCi differ in their sample density it is necessary to interpolate new sample values between those in a given scale SC3, SC2, SCi before that scale SC3, SC2, SCi is subtracted from the next-lower scale SC2, SCi, SCo. Interpolation may be achieved by expanding the corresponding images I3, E, Ii. In detail, the images I3, I2, Ii may be expanded by doubling the sizes of the images I3, E, Ii with each iteration as indicated by corresponding arrows in Fig. 16. Then the images are subtracted to obtain the bandpass images of the lowpass pyramid as shown in Fig. 16. Since each value of this bandpass pyramid may be obtained by convolving a difference of two Gaussians with the original image Io resembling the Laplacian operators, the bandpass pyramid is also referred to as a Laplacian pyramid LPP The Laplacian pyramid LPP provides the difference image of the blurred versions between each scale SC3, SC2, SCI of the Gaussian pyramid GSP.

[0185] Methods applying these pyramid approaches as described above may be applied to various computer visionbased tasks. Thereby, different combinations of the pyramids GSP, LPP for multi-scale analysis may be possible, which are also implied therein. In the following, the detailed explanation of a dynamic scale pyramid DSP according to the present disclosure will be made based on the principles and concepts of the Gaussian / Laplacian pyramids GSP and LPP shown in Figs. 15 and 16 and described in the previous paragraphs.

[0186] The method for performing a computer vision-based task according to the present disclosure comprises the following operations.

[0187] In Operation S201, an input image Io is received. The input image Io may comprise image data obtained by an image sensor 310, which will be explained in more detail with reference to Fig. 23.

[0188] In Operation S202, the input image Io is processed at different resolutions to operate on image data of the input image Io at different scales SCo, SCi, SC2, SC3 by applying a dynamic scale pyramid DSP that provides a variable number of scales SCo, SCi, SC2, SC3 representing the input image Io at different levels of resolution.

[0189] Fig. 17 illustrates schematically a process flow of the method described above.

[0190] Specifically, the provision of a dynamic scale pyramid DSP allows providing additional scales SCi, SC2, SC3 (and their associated images) or skipping scales SCi, SC2, SC3 (and their associated images) on the fly. This results in more efficient performance of a computer vision-based task such as, e.g., object detection or pattern recognition. The dynamic scale pyramid approach differs from conventional ones, which use a fixed number of scales.

[0191] More specifically, in operation S203 as shown in Fig. 17, the number of scales SCi, SC2, SC3 may be dynamical adapted based on information related to the input image Io. The number of scales SCi, SC2, SC3 may be increased or decreased depending on the information. Examples for this dynamical adaption are explained in the following.

[0192] In general, when training a neural network, the different models with different predefined architectures and depth have to be trained for input images having different input resolutions to achieve similar prediction performance. For example, “EfficientDet” (see for example, arXiv: 1911.09070v7, incorporated herein by reference), which refers to a common object detector including a weighted bi-directional feature pyramid network (BiFPN) and a compound scaling method, different depth of the BiFPN need to be trained with different resolution datasets as shown below.

[0193] That is, each network may be trained individually with a different number of fixed scales and depth.

[0194] Fig. 18 is a schematic diagram of an exemplary input image Io with two different resolutions that is to be processed by applying a fixed number of scales SCi, SC2, e.g., two, to train a neural network (not shown in Fig. 18). The left part shows an input image ISD having a standard definition (for example, 512 x 512, 640 x 360 pixels, 720 x 576 pixels and / or 720 x 480 pixels) as input resolution. The right side shows an input image IFHD having a full high-definition (such as, e.g., 1024 x 720 pixels and / or 1920 x 1080 pixels). The shaded area indicates a kernel or filter with a specific kernel size KS or filter size used to process (convolving) the image data of the input image Io at scale SCo providing an image of reduced resolution ft at the subsequent scale SCi. This may represent the first cycle applying a pyramid as described above regarding Figs. 15 and 16. In greater detail, the shaded region may also indicate an input space of the neural network known as receptive field. After processing the scale SCo, the receptive field may be about 75 % for the standard definition input image ISD, and about 37.5 % for the full high-definition input image IFHD.

[0195] Now, when using the full high-definition input image IFHD for a network with fixed scales trained on the standard definition input image ISD, the receptive field and thus prediction performance decreases without access to a dynamic pyramid scaling. That is, the full high-definition input image IFHD may need to be rescaled to match the resolution that the network was originally trained on.

[0196] With the Dynamic scale Pyramid DSP, the network may only need to be trained once and it may run on any (larger) input resolutions of input images Io.

[0197] Figs. 19A and 19B are schematic block diagrams of exemplary neural networks 1000 applying a dynamic scale pyramid DSP. Non-limiting examples of such neural networks 1000 may comprise ConvNets or U- nets.

[0198] In detail, Figs. 19A and 19B show the underlying architecture, where in Fig. 19A low-resolution input images Io, e.g., the standard definition input image ISD, and in Fig. 19B high resolution input images Io, e.g., the full high-definition input image IFHD, serve as input into the network 1000 for training. The cycles for performing the convolution are indicated by AvgPool2D* . AvgPool2D* refers to a pooling layer applied in general ConvNets architectures, which returns an average value from the portion of the image covered by the kernel or filter as shown on the left parts in Figs. 19A and 19B. The network 1000 may further apply encoders / decoders ENC / DEC to use full weight sharing across all scales SCi, SC2, SC3, SC4.

[0199] In greater detail, the information related to the input image Io as described above may comprises metadata related to the input image Io including a specific filter size or kernel size KS for processing pixels of the input image Io and a required receptive field for the computer vision-based task. That is, the dynamic scale pyramid DSP may be initialized with metadata such as a convolution kernel size KS and a maximum desired receptive field in percent of image dimension.

[0200] Then, for each input image Io, a number of scales SCi, SC2, SC3, SC4 may be determined using the corresponding metadata related to the respective input image Io.

[0201] In this context, a default number of scales SCi, SC2, SC3, SC4 may be determined based on metadata related to accuracy or computational requirement.

[0202] The number of scales SCi, SC2, SC3, SC4 may be increased for high-resolution input images Io and decreased for low-resolution input images Io.

[0203] Thus, as illustrated schematically in Figs. 19A and 19B, the number of scales SCi, SC2, SC3, SC4 may be changed dynamically. The number of scales SCi, SC2, SC3 may be increased for processing pixels of the input image Io with a specific filter or kernel if the input image Io has a high-resolution. Furthermore, the number of scales SCi, SC2, SC3 may be decreased for processing the pixels of the input image Io with the specific filter or kernel if the input image Io has a low-resolution. This may allow a flexible training. For example, the low-resolution input images Io is indicated as ED in Fig. 19A. As can be seen, the scale SC4 resulting from a corresponding convolution of Image I3 may not be required and may not be considered. This is indicated by small crosses. Moreover, the scales SCi, SC2, SC3 may be dynamically scaled up for high resolution input images indicated as IFHD in Fig. 19B to cope with large motion areas / objects. As can be seen in Fig. 19B, an additional scale SC4 is considered.

[0204] The dynamical scale pyramid approach may imply a lower training time, cost (energy consumption) and computer hardware required.

[0205] For example, in case of a low-resolution input image Io with a resolution of 512 x 512 pixels may require a training time of about 5 h and a GPU memory of 3GB.

[0206] According to a further example, in case of a high-resolution input image Io with a resolution of 1024 x 1024 pixels may require a training time of about 20h and a GPU memory of 12GB.

[0207] To summarize, using the dynamical scale pyramid DSP, the network 1000 nay not need to be trained on a fixed number of scales SCi, SC2, SC3, SC4and does not require re-training when moving to larger-re solution input. Therefore, applying the dynamical scale pyramid DSP may allow a greater flexibility during training and evaluation. It may train on a smaller dataset with a lower number of scales SCi, SC2, SC3 and dynamically scale up to high-resolution input for a better receptive field and large motion estimation.

[0208] In addition to the example above, the dynamical scale pyramid DSP may also improve the performance in computer vision-based tasks by applying information such as event data. Thus, the information related to the input image Io may comprise event-based data Eo related to the input image Io. Fig. 20 is a schematic block diagram for performing a computer vision-based task applying an exemplary dynamic scale pyramid-based method as described above, which considers information in form of event data. The event-based data Eomay be obtained by an event-based vision sensor 330 (for example, an EVS), which will be explained in reference to Fig. 23.

[0209] The underlying architecture comprises an encoder / decoder structure ENC, DEC. The input comprises an input image Io and corresponding event-based data Eo related to the input image Io. In the first cycle, the input image Io and the corresponding event-based data Eomay be down-sampled as indicated in Fig. 20 by DSAMP.

[0210] Subsequently, for a given scale, e.g., the first scale SCi in the present example, the occurrence of events from event-based data Eo related to the input image Io may be determined. This refers to operation S204 as shown in Fig. 17.

[0211] If the occurrence of events is above a pre-defined threshold THR, a further scale SC2 may be provided, and the subsequent cycle may be performed. This refers to operation S205a in Fig. 17.

[0212] If the occurrence of events is below the pre-defined threshold THR, further scales SC2, SC3, SC4inay be skipped. This is indicated by small crosses in Fig. 20 regarding the second scale SC2 for which the occurrence of events is below the threshold THR and thus, no further cycles may be performed and corresponding scales SC3, SC4 may not be considered. This refers to operation S205b in Fig. 17.

[0213] According to embodiments, the input for determining whether the occurrence of events is above or below the threshold THR may be events-only (sparse) or events concatenated with image data (dense).

[0214] By leveraging sparse event information, it may be possible to determine whether an additional scale SCi, SC2, SC3, SC4 is required or not. This may be based by analyzing motion information from the sparse events.

[0215] In this context, the threshold THR may represent a novelty heuristic that may comprise a threshold for, e.g., the amount of perceived motion in a scene or another metric correlated with whether a downstream task needs to be triggered or not.

[0216] A classifier (not shown in Fig. 20) may then decide whether enough novelty or entropy is available and may trigger the next scale SCi, SC2, SC3, SC4.

[0217] The classifier may also decide to skip the rest of the scales SCi, SC2, SC3, SC4, when trained end-to-end to infer best performance to cost ratio. This is also applicable to neural networks such as U-Nets.

[0218] In addition, the method may also apply a scale-invariant feature transform (SIFT), which is a computer vision algorithm to detect, describe, and match local features in images. More specifically, the SIFT may detect resilient features and local extrema across various scales SCo, SCi, SC2, SC3, SC4. Assuming a default number of four scales SCi, SC2, SC3, SC4. the SIFT may leverage events from the event-based vision sensor 330 such that the number of required scales SCi, SC2, SC3, SC4inay be determined beforehand using the novelty heuristic provided by the threshold THR.

[0219] For example, in case of few events (e.g., a little number of edges or low contrast), there may be a lower chance of finding sufficient features and thus several scales SCi, SC2, SC3, SC4 inay be skipped to be computationally more efficient.

[0220] In case of a larger number of events, there may be a higher chance of finding good keypoints, thus more scales SCi, SC2, SC3, SC4inay be employed.

[0221] To summarize, the scales SCo, SCi, SC2 ofthe dynamic scale pyramid DSP may be dynamically adapted, e.g., additional scales SC3, SC4 may be added, or scales SC2, SC3, SC4 may be skipped. It is noted that the number of scales shown in Fig. 20 is only exemplary. Further or fewer scales SCi, SC2, SC3, SC4 may be applied depending on the information.

[0222] Fig. 21 is a schematic block diagram for performing a computer vision-based task applying a further exemplary dynamic scale pyramid-based method. In this example, the number of the scales SCo, SCi, SC2 to be processed may be determined depending on the event-based data Eo related to the input image Io. In detail, the order of the dynamic scale pyramid DSP may be reversed, as the highest input resolution requires the largest computational effort. Thus, by analyzing event data Eo first, it may be determined whether applying an algorithm / encoder / task on the lowest scale (e.g., SC2 in this example) is enough or it is required to move up to a higher scales SCi, SCo for more accuracy.

[0223] For example, as shown in Fig. 21, applying the threshold THR as described with regard to Fig. 20, it may be determined on basis of the occurrences of events that performing image processing on the scales SC2 and SCi is sufficient. Therefore, the scale SCo having as input the input image Io with the highest resolution may not need to be considered. This is indicated by small crosses in Fig. 21.

[0224] Fig. 22 is a schematic block diagram for performing a computer vision-based task applying a further exemplary dynamic scale pyramid-based method. In this example, the input image Io may be divided into blocks of pixels, and a scale pyramid DSP may be created for each block of the input image Io. More specifically, a given block may be divided into additional blocks as illustrated in Fig. 22 having a different size depending on the event-based data Eo related to the input image Io. The order of processing the blocks of different sizes may be determined depending on the event-based data Eo.

[0225] The example shown in Fig. 22 may present a use-case, for example, for a blocking mechanism provided in e.g., video encoding (c.f. HEVC encoder). In this case, the order of the block sizes may also be reversed, starting from lowest resolution (largest block size) as the highest input resolution requires the largest computational effort. Thus, by analyzing the event based data Eo first, it may be determined whether applying an algorithm / encoder / task on the lowest scale (e.g., SC2) is sufficient or whether it is required to move up to a finer grid for more accuracy.

[0226] Fig. 23 is a schematic block diagram of an exemplary apparatus 300. The apparatus may comprise the image sensor 320, the event-based vision sensor 330, and a processing unit 340 configured to execute the method described above. Non-limiting examples of an image sensor 320 may include but are not limited to a charge- coupled device (CCD) and an active-pixel sensor (CMOS sensor), which represent digital sensors and are applied as e.g., color sensors such as high-quality RGB sensors.

[0227] In addition, the event-based vision sensor 330 may be a conventional event based / dynamic vision sensor (EVS / DVS) as described therein.

[0228] The processing unit 340 may be configured as a central processing unit (CPU) as commonly used in image processing.

[0229] The technology according to the above (i.e. the present technology) is applicable to various products. For example, the technology according to the present disclosure may be realized as a device that is installed on any kind of moving bodies, for example, vehicles, electric vehicles, hybrid electric vehicles, motorcycles, bicycles, personal mobilities, airplanes, drones, ships, and robots.

[0230] Fig. 24 is a block diagram depicting an example of schematic configuration of a vehicle control system as an example of a mobile body control system to which the technology according to an embodiment of the present disclosure can be applied.

[0231] The vehicle control system 12000 includes a plurality of electronic control units connected to each other via a communication network 12001. In the example depicted in Fig. 24, the vehicle control system 12000 includes a driving system control unit 12010, a body system control unit 12020, an outside-vehicle information detecting unit 12030, an in-vehicle information detecting unit 12040, and an integrated control unit 12050. In addition, a microcomputer 12051, a sound / image output section 12052, and a vehiclemounted network interface (I / F) 12053 are illustrated as a functional configuration of the integrated control unit 12050.

[0232] The driving system control unit 12010 controls the operation of devices related to the driving system of the vehicle in accordance with various kinds of programs. For example, the driving system control unit 12010 functions as a control device for a driving force generating device for generating the driving force of the vehicle, such as an internal combustion engine, a driving motor, or the like, a driving force transmitting mechanism for transmitting the driving force to wheels, a steering mechanism for adjusting the steering angle of the vehicle, a braking device for generating the braking force of the vehicle, and the like.

[0233] The body system control unit 12020 controls the operation of various kinds of devices provided to a vehicle body in accordance with various kinds of programs. For example, the body system control unit 12020 functions as a control device for a keyless entry system, a smart key system, a power window device, or various kinds of lamps such as a headlamp, a backup lamp, a brake lamp, a turn signal, a fog lamp, or the like. In this case, radio waves transmitted from a mobile device as an alternative to a key or signals of various kinds of switches can be input to the body system control unit 12020. The body system control unit 12020 receives these input radio waves or signals, and controls a door lock device, the power window device, the lamps, or the like of the vehicle.

[0234] The outside-vehicle information detecting unit 12030 detects information about the outside of the vehicle including the vehicle control system 12000. For example, the outside-vehicle information detecting unit 12030 is connected with an imaging section 12031. The outside-vehicle information detecting unit 12030 makes the imaging section 12031 image an image of the outside of the vehicle, and receives the imaged image. On the basis of the received image, the outside-vehicle information detecting unit 12030 may perform processing of detecting an object such as a human, a vehicle, an obstacle, a sign, a character on a road surface, or the like, or processing of detecting a distance thereto.

[0235] The imaging section 12031 is an optical sensor that receives light, and which outputs an electric signal corresponding to a received light amount of the light. The imaging section 12031 can output the electric signal as an image, or can output the electric signal as information about a measured distance. In addition, the light received by the imaging section 12031 may be visible light, or may be invisible light such as infrared rays or the like.

[0236] The in-vehicle information detecting unit 12040 detects information about the inside of the vehicle. The in-vehicle information detecting unit 12040 is, for example, connected with a driver state detecting section 12041 that detects the state of a driver. The driver state detecting section 12041, for example, includes a camera that images the driver. On the basis of detection information input from the driver state detecting section 12041, the in-vehicle information detecting unit 12040 may calculate a degree of fatigue of the driver or a degree of concentration of the driver, or may determine whether the driver is dozing.

[0237] The microcomputer 12051 can calculate a control target value for the driving force generating device, the steering mechanism, or the braking device on the basis of the information about the inside or outside of the vehicle which information is obtained by the outside-vehicle information detecting unit 12030 or the in- vehicle information detecting unit 12040, and output a control command to the driving system control unit 12010. For example, the microcomputer 12051 can perform cooperative control intended to implement functions of an advanced driver assistance system (ADAS) which functions include collision avoidance or shock mitigation for the vehicle, following driving based on a following distance, vehicle speed maintaining driving, a warning of collision of the vehicle, a warning of deviation of the vehicle from a lane, or the like.

[0238] In addition, the microcomputer 12051 can perform cooperative control intended for automatic driving, which makes the vehicle to travel autonomously without depending on the operation of the driver, or the like, by controlling the driving force generating device, the steering mechanism, the braking device, or the like on the basis of the information about the outside or inside of the vehicle which information is obtained by the outside-vehicle information detecting unit 12030 or the in-vehicle information detecting unit 12040. In addition, the microcomputer 12051 can output a control command to the body system control unit 12020 on the basis of the information about the outside of the vehicle which information is obtained by the outside vehicle information detecting unit 12030. For example, the microcomputer 12051 can perform cooperative control intended to prevent a glare by controlling the headlamp so as to change from a high beam to a low beam, for example, in accordance with the position of a preceding vehicle or an oncoming vehicle detected by the outside-vehicle information detecting unit 12030.

[0239] The sound / image output section 12052 transmits an output signal of at least one of a sound and an image to an output device capable of visually or auditorily notifying information to an occupant of the vehicle or the outside of the vehicle. In the example of Fig. 23, an audio speaker 12061, a display section 12062, and an instrument panel 12063 are illustrated as the output device. The display section 12062 may, for example, include at least one of an on-board display and a head-up display.

[0240] Fig. 25 is a diagram depicting an example of the installation position of the imaging section 12031.

[0241] In Fig. 25, the imaging section 12031 includes imaging sections 12101, 12102, 12103, 12104, and 12105.

[0242] The imaging sections 12101, 12102, 12103, 12104, and 12105 are, for example, disposed at positions on a front nose, sideview mirrors, a rear bumper, and a back door of the vehicle 12100 as well as a position on an upper portion of a windshield within the interior of the vehicle. The imaging section 12101 provided to the front nose and the imaging section 12105 provided to the upper portion of the windshield within the interior of the vehicle obtain mainly an image of the front of the vehicle 12100. The imaging sections 12102 and 12103 provided to the sideview mirrors obtain mainly an image of the sides of the vehicle 12100. The imaging section 12104 provided to the rear bumper or the back door obtains mainly an image of the rear of the vehicle 12100. The imaging section 12105 provided to the upper portion of the windshield within the interior of the vehicle is used mainly to detect a preceding vehicle, a pedestrian, an obstacle, a signal, a traffic sign, a lane, or the like.

[0243] Incidentally, Fig. 25 depicts an example of photographing ranges of the imaging sections 12101 to 12104. An imaging range 12111 represents the imaging range of the imaging section 12101 provided to the front nose. Imaging ranges 12112 and 12113 respectively represent the imaging ranges of the imaging sections 12102 and 12103 provided to the sideview mirrors. An imaging range 12114 represents the imaging range of the imaging section 12104 provided to the rear bumper or the back door. A bird’s-eye image of the vehicle

[0244] 12100 as viewed from above is obtained by superimposing image data imaged by the imaging sections

[0245] 12101 to 12104, for example.

[0246] At least one of the imaging sections 12101 to 12104 may have a function of obtaining distance information. For example, at least one of the imaging sections 12101 to 12104 may be a stereo camera constituted of a plurality of imaging elements, or may be an imaging element having pixels for phase difference detection.

[0247] For example, the microcomputer 12051 can determine a distance to each three-dimensional object within the imaging ranges 12111 to 12114 and a temporal change in the distance (relative speed with respect to the vehicle 12100) on the basis of the distance information obtained from the imaging sections 12101 to 12104, and thereby extract, as a preceding vehicle, a nearest three-dimensional object in particular that is present on a traveling path of the vehicle 12100 and which travels in substantially the same direction as the vehicle 12100 at a predetermined speed (for example, equal to or more than 0 km / hour). Further, the microcomputer 12051 can set a following distance to be maintained in front of a preceding vehicle in advance, and perform automatic brake control (including following stop control), automatic acceleration control (including following start control), or the like. It is thus possible to perform cooperative control intended for automatic driving that makes the vehicle travel autonomously without depending on the operation of the driver or the like.

[0248] For example, the microcomputer 12051 can classify three-dimensional object data on three-dimensional objects into three-dimensional object data of a two-wheeled vehicle, a standard-sized vehicle, a large-sized vehicle, a pedestrian, a utility pole, and other three-dimensional objects on the basis of the distance information obtained from the imaging sections 12101 to 12104, extract the classified three-dimensional object data, and use the extracted three-dimensional object data for automatic avoidance of an obstacle. For example, the microcomputer 12051 identifies obstacles around the vehicle 12100 as obstacles that the driver of the vehicle 12100 can recognize visually and obstacles that are difficult for the driver of the vehicle 12100 to recognize visually. Then, the microcomputer 12051 determines a collision risk indicating a risk of collision with each obstacle. In a situation in which the collision risk is equal to or higher than a set value and there is thus a possibility of collision, the microcomputer 12051 outputs a warning to the driver via the audio speaker 12061 or the display section 12062, and performs forced deceleration or avoidance steering via the driving system control unit 12010. The microcomputer 12051 can thereby assist in driving to avoid collision.

[0249] At least one of the imaging sections 12101 to 12104 may be an infrared camera that detects infrared rays. The microcomputer 12051 can, for example, recognize a pedestrian by determining whether or not there is a pedestrian in imaged images of the imaging sections 12101 to 12104. Such recognition of a pedestrian is, for example, performed by a procedure of extracting characteristic points in the imaged images of the imaging sections 12101 to 12104 as infrared cameras and a procedure of determining whether or not it is the pedestrian by performing pattern matching processing on a series of characteristic points representing the contour of the object. When the microcomputer 12051 determines that there is a pedestrian in the imaged images of the imaging sections 12101 to 12104, and thus recognizes the pedestrian, the sound / image output section 12052 controls the display section 12062 so that a square contour line for emphasis is displayed so as to be superimposed on the recognized pedestrian. The sound / image output section 12052 may also control the display section 12062 so that an icon or the like representing the pedestrian is displayed at a desired position.

[0250] An example of the vehicle control system to which the technology according to the present disclosure is applicable has been described above. The technology according to the present disclosure is applicable to the imaging section 12031 among the above-mentioned configurations. Specifically, the sensor device 10 is applicable to the imaging section 12031. The imaging section 12031 to which the technology according to the present disclosure has been applied flexibly acquires event data and performs data processing on the event data, thereby being capable of providing appropriate driving assistance.

[0251] Note that, the embodiments of the present technology are not limited to the above-mentioned embodiment, and various modifications can be made without departing from the gist of the present technology.

[0252] Further, the effects described herein are only exemplary and not limited, and other effects may be provided.

[0253] Note that, the present technology can also take the following configurations.

[0254] [1] A method for performing a computer vision-based task, the method comprising: receiving (S201) an input image (Io); and processing (S202) the input image (Io) at different resolutions to operate on image data of the input image (Io) at different scales (SCo, SCi, SC2, SC3, SC4) by applying a dynamic scale pyramid (DSP) that provides a variable number of scales (SCo, SCI, SC2, SC3, SC4) representing the input image (Io) at different levels of resolution.

[0255] [2] The method according to [1], wherein the method further comprises: adapting (S203) the number of scales (SCi, SC2, SC3, SC4) dynamically based on information related to the input image (Io), wherein the number of scales (SCi, SC2, SC3, SC4) is increased or decreased depending on the information.

[0256] [3] The method according to [1] or [2], wherein a default number of scales (SCi, SC2, SC3, SC4) is determined based on metadata related to accuracy or computational requirements.

[0257] [4] The method according to [2] or [3], wherein the information related to the input image (Io) comprises metadata related to the input image (Io) including a specific filter size (KS) for processing pixels of the input image (Io) and a required receptive field for the computer vision-based task.

[0258] [5] The method according to [4], wherein the number of scales (SCi, SC2, SC3, SC4) is increased for processing pixels of the input image (Io) with the specific filter if the input image (Io) has a high-resolution, and wherein the number of scales (SCi, SC2, SC3, SC4) is decreased for processing the pixels of the input image (Io) with the specific filter if the input image (Io) has a low-resolution.

[0259] [6] The method according to any one of [1] to [5], wherein the dynamic scale pyramid (DSP) comprises a Gaussian scale pyramid.

[0260] [7] The method according to any one of [2] to [3], wherein the information related to the input image (Io) comprises event-based data (Eo) related to the input image (Io).

[0261] [8] The method according to [7], further comprising: determining (S204), for a given scale (SCi, SC2, SC3, SC4), the occurrence of events from eventbased data (Eo) related to the input image (Io); providing (S205a) a further scale (SCi, SC2, SC3, SC4) if the occurrence of events is above a predefined threshold (THR); or skipping further scales (SCi, SC2, SC3, SC4), if the occurrence of events is below the pre-defined threshold (THR).

[0262] [9] The method according to [7] or [8], wherein the number of the scales (SCi, SC2, SC3, SC4) to be processed is determined depending on event-based data (Eo) related to the input image (Io).

[0263]

[0010] The method according to any one of [7] to [9], wherein the input image (Io) is divided into blocks of pixels, and a dynamic scale pyramid (DSP) is created for each block of the input image (Io), wherein a given block is divided into additional blocks having a different size depending on event-based data (Eo) related to the input image (Io), and wherein the order of processing the blocks of different sizes is determined depending on the event-based data (Eo).

[0264]

[0011] A neural network (1000) configured to apply the method according to any one of [1] to

[0010] ,

[0265]

[0012] The neural network (1000) according to

[0011] , wherein, for each input image (Io), a number of scales (SCi, SC2, SC3, SC4) is determined using corresponding metadata related to the respective input image (Io).

[0266]

[0013] The neural network (1000) according to

[0011] or

[0012] , wherein the number of scales (SCi, SC2, SC3, SC4) is increased for high-resolution input images (Io) and decreased for low-resolution input images (Io).

[0267]

[0014] The neural network (1000) according to any one of

[0011] to

[0013] , wherein the neural network (1000) is a convolutional neural network (ConvNet) providing weight sharing across all scales (SCo, SCi, SC2, SC3, SC4).

[0268]

[0015] An apparatus (300) comprising an image sensor (320), an event-based vision sensor (330), and a processing unit (340) configured to execute the method according to any one of [1] to

[0010] ,

Claims

CLAIMS1. A method for performing a computer vision-based task, the method comprising: receiving an input image; and processing the input image at different resolutions to operate on image data of the input image at different scales by applying a dynamic scale pyramid that provides a variable number of scales representing the input image at different levels of resolution.

2. The method according to claim 1, wherein the method further comprises: adapting the number of scales dynamically based on information related to the input image, wherein the number of scales is increased or decreased depending on the information.

3. The method according to claim 1, wherein a default number of scales is determined based on metadata related to accuracy or computational requirements.

4. The method according to claim 2, wherein information related to the input image comprises metadata related to the input image including a specific filter size for processing pixels of the input image and a required receptive field for the computer vision-based task.

5. The method according to claim 4, wherein the number of scales is increased for processing pixels of the input image with the specific filter if the input image has a high-resolution, and wherein the number of scales is decreased for processing the pixels of the input image with the specific filter if the input image has a low-resolution.

6. The method according to claim 1, wherein the dynamic scale pyramid comprises a Gaussian scale pyramid.

7. The method according to claim 2, wherein the information related to the input image comprises event-based data related to the input image.

8. The method according to claim 7, further comprising: determining, for a given scale, the occurrence of events from event-based data related to the input image; providing a further scale, if the occurrence of events is above a pre-defined threshold; or skipping further scales, if the occurrence of events is below the pre-defined threshold.

9. The method according to claim 7, wherein the number of the scales to be processed is determined depending on event-based data related to the input image.

10. The method according to claim 7, wherein the input image is divided into blocks of pixels, and a dynamic scale pyramid is created for each block of the input image, wherein a given block is divided into additional blocks having a different size depending on event-based data related to the input image, andwherein the order of processing the blocks of different sizes is determined depending on the event-based data.

11. A neural network configured to apply a method according to claim 1.

12. The neural network according to claim 11, wherein, for each input image, a number of scales is determined using corresponding metadata related to the respective input image.

13. The neural network according to claim 11, wherein the number of scales is increased for high- resolution input images and decreased for low-resolution input images.

14. The neural network according to claim 11, wherein the neural network is a convolutional neural network providing weight sharing across all scales.

15. An apparatus comprising an image sensor, an event-based vision sensor, and a processing unit configured to execute a method according to claim 1.

Images