Depth information in inspection images

Abstract

In an example, it is disclosed apparatus for inspecting cargo, the apparatus comprising: one or more detector arrays, each detector array comprising a plurality of detectors, each detector being configured to sense and output detector data indicative of a level of transmission of inspection radiation through the cargo under inspection; a memory that stores a trained calculator configured to calculate a depth information indicative of a distance of one or more objects in the cargo, from a source of the inspection radiation; and a controller configured to: process the detector data from the one or more detector arrays, and use the stored trained calculator to calculate the depth information for one or more objects in the cargo, based on the processed the detector data.

Description

[0001] DEPTH INFORMATION IN INSPECTION IMAGES

[0002] Field of Invention

[0003] The invention relates but is not limited to apparatus for inspecting cargo, e.g., for enhancing and / or assisting detection of one or more objects of interest in an inspection image of cargo. The invention also relates but is not limited to a method for training a calculator, e.g., for use in detection of one or more objects of interest in cargo inspection images. The invention also relates but is not limited to producing corresponding apparatus. The invention also relates but is not limited to corresponding methods, apparatuses and computer programs or computer program products.

[0004] Background

[0005] Inspection images of cargo under inspection may be generated using penetrating radiation, such as X-rays. These images allow detection of objects of interest, such as weapons or contraband as non-limiting examples, within the cargo.

[0006] However, given the structure of the inspection apparatus and the use of penetrating radiation, the penetrating radiation is projected on detectors after transmission through the cargo, resulting in 2D inspection images of the cargo under inspection. Detection of objects of interest may be difficult, and, in ambiguous cases, manual inspection of the cargo may be required, slowing a screening process by an operator.

[0007] Summary of Invention

[0008] Aspects and embodiments of the invention are set out in the appended claims. These and other aspects of the invention, and aspects and embodiments which are useful in understanding the invention set out in the appended claims, are also described in the disclosure herein.

[0009] Any feature in one aspect of the disclosure may be applied to other aspects of the disclosure, in any appropriate combination. In particular, method aspects may be applied to apparatus and computer program aspects, and vice versa.

[0010] Furthermore, features implemented in hardware may generally be implemented in software, and vice versa. Any reference to software and hardware features herein should be construed accordingly. Brief Description of Drawings

[0011] Embodiments of the present disclosure will now be described, by way of example, with reference to the accompanying drawings, in which:

[0012] Figure 1 schematically illustrates an example apparatus according to the disclosure;

[0013] Figure 2 shows a flow chart illustrating an example method according to the disclosure;

[0014] Figure 3A illustrates an example apparatus according to the disclosure, in operation;

[0015] Figure 3B illustrates an example apparatus according to the disclosure, in operation;

[0016] Figure 3C schematically illustrates an example of a depth image according to the disclosure;

[0017] Figure 4 shows a flow chart illustrating another example method according to the disclosure; and

[0018] Figure 5 shows a flow chart illustrating another example method according to the disclosure.

[0019] In the figures, similar elements bear identical numerical references.

[0020] Description of Example Embodiments

[0021] Applying a trained calculator to calculate depth information

[0022] As illustrated in Figure 1 , the disclosure discloses an example apparatus 15 configured to detect one or more objects of interest in an inspection image of cargo, the cargo being referred to as 155 in Figure 1.

[0023] An example of one or more inspection images 10 (e.g., inspection images 10-1 , 10-2, 10- 3, 10-4 and 10-5) is schematically represented in Figures 3A and 3B.

[0024] As illustrated in Figure 1 , the inspection images 10 are generated using penetrating radiation 2 transmitted from a source 156 of the penetrating radiation 2 of an inspection system 154. The penetrating radiation 2 is transmitted through the cargo 155 under inspection and projected on one or more detector arrays 3 of the apparatus 15.

[0025] Each detector array 3 comprises a plurality of detectors. Each detector is configured to sense and output detector data indicative of a level of transmission of the inspection radiation 2 through the cargo 155 under inspection.

[0026] In some examples, the one or more detector arrays 3 comprise a matrix of detectors, such that, in some examples, the matrix of detectors comprises five detector arrays. Other numbers of detector arrays are envisaged. In Figures 3A and 3B, the detector arrays 3 comprise a matrix of detectors comprising five detector arrays, and inspection images 10- 1 , 10-2, 10-3, 10-4 and 10-5 are generated, based on the processed detector data from the plurality of detectors corresponding to each one of the five detector arrays, respectively.

[0027] In some examples (not shown in the Figures), the one or more detector arrays comprise a single detector array, such that, in some examples, the single detector array comprises a linear array of detectors.

[0028] Because of the projection of the radiation 2 on the one or more detector arrays 3, the inspection images 10 are 2D images.

[0029] In the present disclosure, in order to assist the detection of one or more objects of interest in the inspection images of the cargo 155, depth information is calculated for one or more objects in the cargo 155, as explained in greater detail below.

[0030] In Figure 1 , the apparatus 15 comprises a memory 151 storing a trained calculator 1 (a deep learning algorithm) configured to calculate the depth information indicative of a distance of the one or more objects in the cargo 155, from the source 156 of the inspection radiation 2. The apparatus 15 also comprises a controller 152 coupled to the memory 151. The controller 152 is configured to, in operation, perform the method 100 as illustrated in Figure 2.

[0031] The method 100 of Figure 2 comprises: processing, at S1 , the detector data from the one or more detector arrays 3, and using, at S3, the stored trained classifier 1 to calculate the depth information for one or more objects in the cargo, based on the processed the detector data.

[0032] The method 100 of Figure 2 comprises an optional step of generating, at S2, an inspection image 10 of the cargo 155, based on the processed detector data from the plurality of detectors (as illustrated in Figures 3A and 3B). In Figure 3B, an inspection image 10-6 is generated, based on the processed detector data from the plurality of detectors corresponding to inspection images 10-1 , 10-2, 10-3, 10-4 and 10-5 generated by a matrix of five detector arrays.

[0033] Alternatively or additionally, the method 100 of Figure 2 comprises an optional step of generating, at S4, a depth image 101 (as illustrated in Figures 3A and 3B) of the cargo, based on the calculated depth information.

[0034] Therefore, the calculator 1 of the disclosure, after it has been trained, calculates a depth information and enables more accurate detection of objects of interest, such as weapons or contraband products as non-limiting example, e.g., improving safety.

[0035] The calculator 1 of the disclosure, after it has been trained, reduces the need for manual inspection of the cargo, thus accelerating a screening process by an operator.

[0036] The calculator 1 of the disclosure, after it has been trained, calculates the depth information faster than a processor using a conventional method for calculating a depth information, e.g., using a stereoscopic method with a matrix of detectors (also called “computer vision” method). An example of a method using a computer vision method is disclosed in GB2589926. The calculation time using the method of the disclosure is 1 / 3- 1 / 2 of the conventional calculation time using a conventional computer vision method. Calculation of a depth information is not possible on conventional apparatus comprising a single detector array 3. However, the calculator 1 of the disclosure, after it has been trained, enables calculation of the depth information, even when the apparatus 15 comprise a single detector array 3.

[0037] The calculator 1 of the disclosure, after it has been trained, enables reduction of noisy results and enables the depth information on each object contours to be more homogeneous than with a conventional method calculating a depth information using a computer vision method with a matrix of detectors.

[0038] In Figure 1 , the apparatus 15 comprises a user interface 157 comprising a display. The controller 152 may be further configured to cause the display to display the calculated depth information. In an example, one or more objects which are calculated as being closer to the source 156 of the penetrating radiation than a first threshold are displayed in a first colour (such as red, but other colours may be envisaged), and the one or more objects which are calculated as being further from the source 156 of the penetrating radiation than the first threshold are displayed in a second colour (such as blue, but other colours may be envisaged), different from the first colour.

[0039] As illustrated in Figures 3A and 3B, after the apparatus 15 has been configured with the calculator 1 , the apparatus 15 can use the calculator 1 based on locally acquired inspection images 10 to calculate the depth information.

[0040] The calculator 1 is configured to calculate the depth information based on the inspection images 10, the inspection images 10 comprising one or more features at least similar to the training images used to generate the calculator 1 using a deep learning algorithm, as described in greater detail below.

[0041] In general, the calculator 1 is configured to calculate the depth information of the one or more objects detected in the inspection images 10, in a way similar to the calculation of the depth information performed during the training described in greater detail below. In some examples, the disclosure may be applied for inspection of a real container containing the cargo 155, and at least some of the methods of the disclosure may comprise obtaining inspection images by irradiating, using penetrating radiation, one or more real containers configured to contain the cargo 155, and detecting radiation from the irradiated one or more real containers.

[0042] Generation of the trained calculator

[0043] The disclosure also discloses an example method for generating a trained calculator 1 configured to calculate a depth information indicative of a depth of an object detected in an inspection image of cargo, as calculated from the source 156 of the radiation.

[0044] Figure 4 shows a flow chart illustrating an example method 200 according to the disclosure. The method 200 is for generating a trained calculator configured to calculate a depth information indicative of a distance of one or more objects in the cargo, from a source of the inspection radiation.

[0045] In Figure 4, the method 200 comprises: obtaining, at S21 , a plurality of annotated training streams of detector data, wherein the annotation indicates a ground truth depth for each training stream of detector data; and training, at S22, the calculator to calculate the depth information based on the obtained training streams of detector data, by teaching a machine learning algorithm to minimize a difference between a depth prediction made by the machine learning algorithm during training and the corresponding annotation for each training stream of detector data.

[0046] The plurality of annotated training streams of data comprises streams of data indicative of detector data from one or more detector arrays, each detector array comprising a plurality of detectors. In other words, the deep learning algorithm is trained using a plurality of training streams which are similar to inspection images and comprising one or more objects of interest. The trained calculator 1 is obtained after training.

[0047] In the method 200 of Figure 4, the training streams are annotated. In other words, in the training streams, the values of the depth information of the one or more object in the training streams are known. In some examples, the ground truth (i.e., the known depth information) may be calculated using a conventional computer vision method, from the training streams. Alternatively or additionally, in some examples, a domain specialist (such as a human operator) may manually annotate the training streams with ground truth annotation (inputting ground truth and annotations by an operator may use a man / machine interface).

[0048] The trained calculator 1 is generated based on the training streams obtained at S21. During the training, the number of streams associated with the detector arrays corresponds to the number of detector arrays in the apparatus 15, after the apparatus 15 has been configured with the trained calculator 1 .

[0049] In one example of training, the training streams obtained at S21 correspond to streams of a plurality of detector arrays 3 of a matrix of detectors, such that, in some examples, the matrix of detectors comprises five detector arrays (other numbers are envisaged). Such an example of training is destined to train a calculator for configuring an apparatus as shown in Figures 3A and 3B. In such an example of training, the obtained streams of detector data correspond to inspection images 10-1 , 10-2, 10-3, 10-4 and 10-5 of Figures 3A and 3B.

[0050] In another example of training (not shown in the Figures), the training streams obtained at S21 correspond to streams of a detector array which comprises a single detector array (such as a linear array of detectors). Such an example of training is destined to train a calculator for configuring an apparatus with a single detector array or using a single column of a matrix of detectors.

[0051] As explained in more detail below, the training step S22 mainly involves the calculator inferring depth information for one or more objects in the input training streams, by comparing the inference with the ground truth, and encoding the inferred depth information in the form of the trained calculator 1 . The learning process is typically computationally intensive and may involve large volumes of training streams (such as tens, hundreds or thousands of streams). As shown in Figure 1 , the method 200 may be performed at a computer system 110 separate, optionally remote, from the inspection system 154 and / or the apparatus 15.

[0052] Any suitable calculator may be used. As shown in Figures 3A and 3B, after the training, the trained calculator 1 comprises a trained convolutional neural network.

[0053] As shown in Figure 3A, the trained convolutional neural network comprises a network having a structure called “U-Net”. Other types of convolutional neural networks are envisaged, such as a network having a structure called a “Masked-attention Mask Transformer” (also known as “Mask2Former”) and / or a network having a structure called “SegFormer”.

[0054] The U-Net convolutional neural network of Figure 3A comprises a contracting path 1001 and an expansive path 1002.

[0055] The contracting path 1001 comprises, from the input streams (five streams in the example of Figure 3A - but one single input stream is also envisaged, as explained above), four operations max pooling 2*2, going down the contracting path 1001. The expansive path 1002 comprises four operations up convolution 2*2, going up the expansive path 1002. The layers of each level of the contracting path 1001 and of the expansive path 1002 are separated by convolution 3*3 operations, except just before the output, where a convolution 1*1 operation is used. The levels of the paths 1001 and 1002 of the U-Net architecture are linked by copy and crop operations. The output is the depth information, as shown in the depth image 101 in Figures 3A.

[0056] As already stated, the display of the user interface 157 of the apparatus 15 may be configured to display the calculated depth information.

[0057] A more detailed example of a depth image 101 is also shown at Figure 3C, for clarity. In the examples of Figures 3A, 3B and 3C, the one or more objects represented with a specific colour (other colours are envisaged). In the example of Figures 3A, 3B and 3C, one or more objects which are calculated as being closer to the source of the penetrating radiation than a first threshold are displayed in red, and one or more objects which are calculated as being further from the source of the penetrating radiation than the first threshold but closer than a second threshold are displayed in green; one or more objects which are calculated as being further from the source of the penetrating radiation than the second threshold are displayed in blue.

[0058] Alternatively or additionally, other ways of outputting the calculated depth information are also envisaged, such as by displaying a list of the calculated depth values.

[0059] As already stated, training the calculator 1 comprises training the algorithm to minimize a difference between a depth prediction made by the machine learning algorithm during training and the corresponding annotation for each training stream of detector data.

[0060] As already stated, after it has been trained, the calculator is used for determining the depth information. The learned calculation function may be used to enhance and / or assist detection of objects of interest (e.g., weapons as non-limiting examples) that human operators (e.g. operators in customs organisations) may find difficult to detect in an inspection image.

[0061] As explained in greater detail later with reference to Figure 1 , the apparatus 15 is configurable by the method 200 to generate the calculator 1 configured to calculate the depth information.

[0062] Computer system and detection system

[0063] Figure 1 also schematically illustrates an example computer system 110. Both the computer system 110 and the apparatus 15 are configured to implement, at least partly, the example method 200 of Figure 4. In particular, in a preferred embodiment, the computer system 110 executes the machine learning to generate the calculator 1 to be stored on the apparatus 15. Although a single apparatus 15 is shown for clarity, the computer system 110 may communicate and interact with multiple such apparatuses. The training streams may themselves be obtained using images acquired using the inspection system 154 and / or using other, similar systems and / or using other sensors and data sources. In some examples, the training streams may have been obtained in a different environment, e.g. using a similar systems (or equivalent set of sensors) installed in a different (but preferably similar) environment, or in a controlled test configuration in a laboratory environment.

[0064] The computer system 110 of Figure 1 conventionally comprises a memory 121 , a processor 122 and a communications interface 123. The system 110 may be configured to communicate with one or more apparatuses 15, via the interface 123 and a link 30 (e.g. Wi-Fi connectivity, but other types of connectivity may be envisaged). The memory 121 is configured to store, at least partly, data, for example for use by the processor 122. In some examples the data stored on the memory 121 may comprise data such as the training streams (and the data used to generate the training streams) and / or the machine learning algorithm.

[0065] The apparatus 15 of Figure 1 also comprises a communications interface 153 (e.g. Wi-Fi connectivity, but other types of connectivity may be envisaged) allowing connection to the interface 123 via the link 30.

[0066] It should be understood that, in order to calculate the depth information, the apparatus 15 may be connected, at least temporarily, to the system 110, and the apparatus 15 may access the memory 121 of the system 110.

[0067] In a non-limiting example, the apparatus 15 may also comprise the inspection system 154, as described in greater detail later. The inspection system 154 may be integrated into the apparatus 15 or connected to other parts of the apparatus 15 by wired or wireless connection. In other words, the inspection system 154 may be used to acquire one or more of the plurality of training streams.

[0068] In some examples, the processor 122 of the system 110 may comprise greater computational power and memory resources than the processor 152 of the apparatus 15. The calculator 1 generation is therefore performed, at least partly, remotely from the apparatus 15, at the computer system 110. However, if sufficient processing power is available locally then the calculator 1 learning could be performed (at least partly) by the processor 152 of the apparatus 15.

[0069] Apparatus manufacture

[0070] As illustrated in Figure 5, a method 300 of producing the apparatus 15 for inspecting cargo.

[0071] In Figure 5, the method 300 comprises: obtaining, at S31 , a calculator 1 generated by the method 200 according to any aspects of the disclosure; and storing, at S32, the obtained calculator 1 in the memory 151 of the apparatus 15.

[0072] The calculator 1 may be created and stored using any suitable representation, for example as a data description comprising data elements specifying calculation conditions and their calculation outputs. Such a data description could be encoded e.g., using XML or using a bespoke binary representation. The data description is then interpreted by the processor 152 running on the apparatus 15 when applying the calculator 1 . Alternatively, the machine learning algorithm may generate the calculator 1 directly as executable code (e.g., machine code, virtual machine byte code or interpretable script). This may be in the form of a code routine that the apparatus 15 can invoke to apply the calculator 1 .

[0073] Regardless of the representation of the calculator 1 , the calculator 1 effectively defines a depth calculation algorithm (comprising a set of rules) based on input data (i.e., the one or more inspection images 10).

[0074] The apparatus 15 may be connected temporarily to the system 110 to transfer the generated calculator 1 (e.g., as a data file or executable code) or the transfer may occur using a storage medium (e.g., memory card). In a preferred approach, the calculator 1 is transferred to the apparatus 15 from the system 110 over the network connection 30 (this could include transmission over the Internet from a central location of the system 110 to a local network where the apparatus 15 is located). The calculator 1 is then installed at the apparatus 15. The calculator 1 could be installed as part of a firmware update of apparatus software, or independently.

[0075] Installation of the calculator 1 may be performed once (e.g., at time of manufacture or installation) or repeatedly (e.g., as a regular update). The latter approach can allow the detection performance of the calculator to be improved over time, as new training images become available.

[0076] Further details and examples

[0077] The disclosure may be advantageous but is not limited to customs and / or security applications.

[0078] The disclosure typically applies to cargo inspection systems (e.g., land, sea or air cargo). The inspection system 154 of Figure 1 , acting as an inspection system, may be configured to inspect the container of the cargo 155, e.g., by transmission of inspection radiation through the container of the cargo 155.

[0079] The container of the cargo 155 configured to contain the cargo may be, as a non-limiting example, placed on a vehicle. In some examples, the vehicle may comprise a trailer configured to carry the container of the cargo 155.

[0080] The inspection system 154 of Figure 1 may comprises a source configured to generate the inspection radiation. The radiation source is configured to cause the inspection of the cargo through the material (usually steel) of walls of the container of the cargo 155, e.g. for detection and / or identification of the cargo. Alternatively or additionally, a part of the inspection radiation may be transmitted through the container of the cargo 155 (the material of the container of the cargo 155 being thus transparent to the radiation), while another part of the radiation may, at least partly, be reflected by the container of the cargo 155 (called “back scatter”).

[0081] In some examples, the inspection system 154 may be mobile and may be transported from a location to another location (the inspection system 154 may comprise an automotive vehicle).

[0082] In the source, electrons are generally accelerated under a voltage comprised between 100keV and 15MeV. In mobile inspection systems, the power of the X-ray source may be e.g., between 100keV and 9.0MeV, typically e.g., 300keV, 2MeV, 3.5MeV, 4MeV, or6MeV, fora steel penetration capacity e.g., between 40mm to 400mm, typically e.g., 300mm (12in).

[0083] In static inspection systems, the power of the X-ray source may be e.g., between 1 MeV and 10MeV, typically e.g., 9MeV, for a steel penetration capacity e.g., between 300mm to 450mm, typically e.g., 410mm (16.1 in).

[0084] In some examples, the source may emit successive X-ray pulses. The pulses may be emitted at a given frequency, comprised between 50 Hz and 1000 Hz, for example approximately 200 Hz.

[0085] According to some examples, the detectors may be mounted on a gantry. The gantry for example forms an inverted “L”. In mobile inspection systems, the gantry may comprise an electro-hydraulic boom (not shown on the Figures) which can operate in a retracted position in a transport mode and in an inspection position. The boom may be operated by hydraulic actuators (such as hydraulic cylinders). In static inspection systems, the gantry may comprise a static structure.

[0086] It should be understood that the inspection radiation source may comprise sources of other penetrating radiation, such as, as non-limiting examples, sources of ionizing radiation, for example gamma rays or neutrons. The inspection radiation source may also comprise sources which are not adapted to be activated by a power supply, such as radioactive sources, such as using Co60 or Cs137. In some examples, the inspection system comprises detectors, such as X-ray detectors, optional gamma and / or neutrons detectors, e.g., adapted to detect the presence of radioactive gamma and / or neutrons emitting materials within the cargo, e.g., simultaneously to the X-ray inspection. In some examples, detectors may be placed to receive the radiation reflected by the container.

[0087] In the context of the present disclosure, the container may be any type of container, such as a holder or a box, etc. The container may thus be, as non-limiting examples a palette (for example a palette of European standard, of US standard or of any other standard) and / or a train wagon and / or a tank and / or a boot of the vehicle and / or a “shipping container” (such as a tank or an ISO container or a non-ISO container or a Unit Load Apparatus (ULD) container).

[0088] In some examples, one or more memory elements (e.g., the memory of one of the processors) can store data used for the operations described herein. This includes the memory element being able to store software, logic, code, or processor instructions that are executed to carry out the activities described in the disclosure.

[0089] A processor can execute any type of instructions associated with the data to achieve the operations detailed herein in the disclosure. In one example, the processor could transform an element or an article (e.g., data) from one state or thing to another state or thing. In another example, the activities outlined herein may be implemented with fixed logic or programmable logic (e.g., software / computer instructions executed by a processor) and the elements identified herein could be some type of a programmable processor, programmable digital logic (e.g., a field programmable gate array (FPGA), an erasable programmable read only memory (EPROM), an electrically erasable programmable read only memory (EEPROM)), an ASIC that includes digital logic, software, code, electronic instructions, flash memory, optical disks, CD-ROMs, DVD ROMs, magnetic or optical cards, other types of machine-readable mediums suitable for storing electronic instructions, or any suitable combination thereof.

[0090] As one possibility, there is provided a computer program, computer program product, or computer readable medium, comprising computer program instructions to cause a programmable computer to carry out any one or more of the methods described herein. In example implementations, at least some portions of the activities related to the processors may be implemented in software. It is appreciated that software components of the present disclosure may, if desired, be implemented in ROM (read only memory) form. The software components may, generally, be implemented in hardware, if desired, using conventional techniques.

[0091] Other variations and modifications of the system will be apparent to the skilled in the art in the context of the present disclosure, and various features described above may have advantages with or without other features described above. The above embodiments are to be understood as illustrative examples, and further embodiments are envisaged. It is to be understood that any feature described in relation to any one embodiment may be used alone, or in combination with other features described, and may also be used in combination with one or more features of any other of the embodiments, or any combination of any other of the embodiments. Furthermore, equivalents and modifications not described above may also be employed without departing from the scope of the invention, which is defined in the accompanying claims.

Claims

CLAIMS1. Apparatus for inspecting cargo, the apparatus comprising: one or more detector arrays, each detector array comprising a plurality of detectors, each detector being configured to sense and output detector data indicative of a level of transmission of inspection radiation through the cargo under inspection; a memory that stores a trained calculator configured to calculate a depth information indicative of a distance of one or more objects in the cargo, from a source of the inspection radiation; and a controller configured to: process the detector data from the one or more detector arrays, and use the stored trained calculator to calculate the depth information for one or more objects in the cargo, based on the processed the detector data.

2. The apparatus of claim 1 , wherein the controller is configured to: generate an inspection image of the cargo, based on the processed detector data from the plurality of detectors, and / or generate a depth image of the cargo, based on the calculated depth information.

3. The apparatus of claim 1 or claim 2, further comprising a user interface comprising a display, and wherein the controller is further configured to cause the display to display the calculated depth information, optionally wherein one or more objects which are calculated as being closer to the source of the penetrating radiation than a first threshold are displayed in a first colour, and wherein the one or more objects which are calculated as being further from the source of the penetrating radiation than the first threshold are displayed in a second colour, different from the first colour.

4. The apparatus of any one of claims 1 to 3, wherein the trained calculator comprises a trained convolutional neural network, optionally wherein the trained convolutional neural network comprises a network having a structured called “U-Net”.

5. The apparatus of any preceding claims, wherein the one or more detector arrays comprise a matrix of detectors and wherein the trained calculator is configured to calculate a depth information for one or more objects in the cargo based on processed detector data from the matrix of detectors, optionally wherein the matrix of detectors comprises five detector arrays; and / or wherein the one or more detector arrays comprise a single detector array, and wherein the trained calculator is configured to calculate a depth information for one or more objects in the cargo based on processed detector data from the single detector array, optionally wherein the single detector array comprises a linear array of detectors.

6. A method for generating a trained calculator configured to calculate a depth information indicative of a distance of one or more objects in the cargo, from a source of the inspection radiation, the method comprising: obtaining a plurality of annotated training streams of detector data, wherein the annotation indicates a ground truth depth for each training stream of detector data; and training the calculator to calculate the depth information based on the obtained training streams of detector data, by teaching a machine learning algorithm to minimize a difference between a depth prediction made by the machine learning algorithm during training and the corresponding annotation for each training stream of detector data, wherein the plurality of annotated training streams of data comprises streams of data indicative of detector data from one or more detector arrays, each detector array comprising a plurality of detectors.

7. The method of the preceding claim, wherein the method comprises training the calculator to calculate the depth information based on a plurality of annotated training streams of data indicative of detector data from a plurality of detector arrays, such as five detector arrays of a matrix of detectors; and / or wherein the method comprises training the calculator to calculate the depth information based on a plurality of annotated training streams of data indicative of detector data from a single detector array, such as a linear array of detectors.

8. The method according to claim 6 or claim 7, performed at a computer system separate, optionally remote, from a device configured to inspect cargo.

9. A method of producing apparatus for inspecting cargo, wherein the apparatus comprises one or more detector arrays, each detector array comprising a plurality of detectors, each detector being configured to sense and output detector data indicative of a level of transmission of inspection radiation through the cargo under inspection; a memory; and a controller, wherein the method comprises: obtaining a trained calculator, generated by the method according to any one of claims 6 to 8; and storing the obtained trained calculator in the memory of the apparatus.

10. The method of claim 9, wherein the storing comprises transmitting the generated trained calculator to the apparatus via a network, the apparatus receiving and storing the trained calculator.

11. The method of claim 10, wherein the trained calculator is generated, stored and / or transmitted in the form of one or more of: a data representation of the trained calculator; executable code for applying the trained calculator.

12. A method of processing one or more signals for inspecting cargo, comprising: obtaining detector data indicative of a level of transmission of inspection radiation through cargo under inspection, the detector data being sensed and output by a plurality of detectors of one or more one or more detector arrays; processing the detector data from the one or more detector arrays; and using a trained calculator to calculate a depth information for one or more objects in the cargo under inspection, based on the processed the detector data.

13. The method of claim 12, comprising: generating an inspection image of the cargo, based on the processed detector data from the plurality of detectors, and / orgenerating a depth image of the cargo, based on the calculated depth information.

14. The method of claim 12 or claim 13, comprising using the trained calculator as generated by the method according to any one of claims 6 to 8.

15. A computer program or a computer program product comprising instructions which, when executed by a processor, enable the processor to perform the method according to any one of claims 6 to 8 and / or the method according to any one of claims 9 to 11 and / or the method according to any one of claims 12 to 14.

Images