Image detection device and image detection method

By adjusting and segmenting images, the image detection module can detect objects that are close or far away in the image without increasing resources, solving the problem of high hardware upgrade cost in the prior art and improving detection accuracy and sense of detail .

CN113763456BActive Publication Date: 2024-11-19ALTEK SEMICON
View PDF 0 Cites 0 Cited by

Patent Information

Application Number
CN202010483394.6
Authority / Receiving Office
CN · China
Patent Type
Patents(China)
Current Assignee / Owner
Filing Date
2020-06-01
Publication Date
2024-11-19
Estimated Expiration
2040-06-01

AI Technical Summary

Technical Problem

Existing image detection technologies require adding hardware equipment to improve accuracy, resulting in high costs and large resources consumption, and lack of algorithm optimization methods.

Method used

The image processing module adjusts the image size to generate adjusted images and segmented images. Using these images as input, the image detection module can detect objects that are close or far away in the image, avoiding the increase in memory, frequency and power consumption.

Benefits of technology

It realizes detecting objects with far distances in the image without increasing memory, frequency or power consumption, improving the accuracy and sense of detail of image detection and reducing the need for hardware upgrades.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN113763456B_ABST
    Figure CN113763456B_ABST
Patent Text Reader

Abstract

The present invention provides an image detection device and an image detection method. The image detection method comprises: obtaining an image, wherein the image comprises an object; adjusting a first size of the image to generate an adjusted image; generating a first segmented image and a second segmented image according to the image; and detecting the object in the image according to a plurality of input images, wherein the plurality of input images comprises the first segmented image, the second segmented image and the adjusted image.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The invention relates to an image detection device and an image detection method. Background Art

[0002] With the advancement of technology, image detection technology has been widely used in different fields, such as quality management, traffic management or face detection. The results of image detection are often related to the computing power used for image processing and the resolution of the image. To improve the accuracy of image detection, users often need to purchase more hardware equipment, such as a camera with higher resolution or a higher capacity memory.

[0003] However, improving the accuracy of image detection by upgrading hardware equipment will consume a lot of costs. Therefore, how to propose a method to improve the accuracy of image detection by using an algorithm is one of the goals that people in this field are committed to. Summary of the invention

[0004] The present invention provides an image detection device and an image detection method, which can detect objects at a long distance in an image without increasing memory usage, frequency or power consumption.

[0005] An image detection device of the present invention includes a processor, a storage medium and a transceiver. The storage medium stores multiple modules. The processor is coupled to the storage medium and the transceiver, and accesses and executes multiple modules, wherein the multiple modules include a data collection module, an image processing module and an image detection module. The data collection module obtains an image through the transceiver, wherein the image includes an object. The image processing module adjusts a first size of the image to generate an adjusted image, and generates a first segmented image and a second segmented image based on the image. The image detection module detects objects in the image based on multiple input images, wherein the multiple input images include a first segmented image, a second segmented image and an adjusted image.

[0006] In an embodiment of the present invention, the image processing module segments a first frame of the image to generate a first segmented image, and segments a second frame of the image to generate a second segmented image, wherein the first frame is different from the second frame.

[0007] In an embodiment of the present invention, the image processing module segments a first frame of the image to generate a first segmented image, and adjusts a first size of a third frame of the image to generate an adjusted image, wherein the first frame is different from the third frame.

[0008] In an embodiment of the present invention, the plurality of input images correspond to different frames respectively, wherein the image detection module detects an object in a first frame of the image according to at least one first candidate frame corresponding to a first frame of the plurality of input images.

[0009] In an embodiment of the present invention, the image detection module detects an object in a second frame of the image according to at least one first candidate frame and at least one second candidate frame corresponding to a second frame of the plurality of input images.

[0010] In an embodiment of the present invention, the image detection module determines a target frame according to at least one first candidate frame, and detects an object in a second frame of the image according to the target frame and at least one second candidate frame corresponding to a second frame of the plurality of input images.

[0011] In an embodiment of the present invention, the image processing module adjusts a first size of an image to generate a second adjusted image, and segments the second adjusted image to generate a first segmented image and a second segmented image.

[0012] In one embodiment of the present invention, the above-mentioned image processing module adjusts the first size of the image according to a reference image to generate a third adjusted image, segments the third adjusted image according to the size of a reference object corresponding to the object in the reference image to generate a third segmented image, and adjusts the second size of the third segmented image to generate the first segmented image, wherein the first aspect ratio of the reference image is the same as the second aspect ratio of any one of the multiple input images.

[0013] An image detection method of the present invention includes: obtaining an image, wherein the image includes an object; adjusting a first size of the image to generate an adjusted image; generating a first segmented image and a second segmented image based on the image; and detecting the object in the image based on multiple input images, wherein the multiple input images include the first segmented image, the second segmented image, and the adjusted image.

[0014] In one embodiment of the present invention, the step of generating a first segmented image and a second segmented image according to an image includes: segmenting a first frame of the image to generate a first segmented image, and segmenting a second frame of the image to generate a second segmented image, wherein the first frame is different from the second frame.

[0015] In one embodiment of the present invention, the above-mentioned steps of generating a first segmented image and a second segmented image based on an image include: segmenting a first frame of the image to generate a first segmented image, and the step of adjusting a first size of the image to generate an adjusted image includes: adjusting a first size of a third frame of the image to generate an adjusted image, wherein the first frame is different from the third frame.

[0016] In one embodiment of the present invention, the above-mentioned multiple input images correspond to different frames respectively, and the step of detecting objects in the image based on the multiple input images includes: detecting objects in the first frame of the image based on at least one first candidate frame corresponding to the first frame of the multiple input images.

[0017] In one embodiment of the present invention, the above-mentioned step of detecting objects in the image based on multiple input images includes the steps of: detecting objects in the second frame of the image based on at least one first candidate frame and at least one second candidate frame corresponding to the second frame in the multiple input images.

[0018] In one embodiment of the present invention, the above-mentioned step of detecting objects in an image based on multiple input images includes the steps of: determining a target frame based on at least one first candidate frame, and detecting objects in a second frame of the image based on the target frame and at least one second candidate frame corresponding to a second frame in the multiple input images.

[0019] In an embodiment of the present invention, the step of generating the first segmented image and the second segmented image according to the image includes: adjusting a first size of the image to generate a second adjusted image, and segmenting the second adjusted image to generate the first segmented image and the second segmented image.

[0020] In one embodiment of the present invention, the above-mentioned step of generating a first segmented image and a second segmented image based on an image includes: adjusting a first size of the image based on a reference image to generate a third adjusted image; segmenting the third adjusted image based on a size of a reference object corresponding to an object in the reference image to generate a third segmented image; and adjusting a second size of the third segmented image to generate the first segmented image, wherein a first aspect ratio of the reference image is the same as a second aspect ratio of any one of the multiple input images.

[0021] Based on the above, the image detection device of the present invention can use the adjusted image generated by magnifying the image to detect objects that are closer in the image, and can use the segmented image to detect objects that are farther away in the image. Using the segmented image instead of the entire image as the input image when performing image detection will enable the image detection device to detect objects that are farther away in the image (i.e., smaller or less clear objects in the image). BRIEF DESCRIPTION OF THE DRAWINGS

[0022] Figure 1 A schematic diagram of an image detection device is shown according to an embodiment of the present invention.

[0023] Figure 2 A schematic diagram showing generation of an adjusted image according to an embodiment of the present invention is shown.

[0024] Figure 3 A schematic diagram showing a method of generating a segmented image according to an embodiment of the present invention is shown.

[0025] Figure 4A and Figure 4B A schematic diagram showing a method of generating a segmented image according to another embodiment of the present invention is shown.

[0026] Figure 5 A schematic diagram of a reference image is shown according to an embodiment of the present invention.

[0027] Figure 6 A schematic diagram is shown according to an embodiment of the present invention for detecting objects in an image based on candidate bounding boxes (candidate bboxes) of different frames.

[0028] Figure 7 A schematic diagram of detecting an object in an image based on a target bbox and a candidate bbox is shown according to an embodiment of the present invention.

[0029] Figure 8 A schematic diagram of an image detection method is shown according to an embodiment of the present invention.

[0030] Description of Reference Numerals

[0031] 100: image detection device;

[0032] 110: processor;

[0033] 120: storage medium;

[0034] 121: data collection module;

[0035] 122: image processing module;

[0036] 123: image detection module;

[0037] 130: transceiver;

[0038] 200: image;

[0039] 300, 400, 500: adjusted images;

[0040] 410, 420, 510, 520: segmenting an image;

[0041] 530: overlapping area;

[0042] 600, 610, 620, 630, 640, 700, 710, 720, 730, 740, 750: target frame;

[0043] 601, 602, 611, 612, 613, 621, 622, 701, 702, 711, 712, 713, 721, 722: candidate boxes;

[0044] l0, l1, li, lr: length of the image;

[0045] ld: length of the object / length of the overlapping area;

[0046] P1, P2, P3: Process;

[0047] S801, S802, S803, S804: steps;

[0048] w0, w1, wi, wr: image width. DETAILED DESCRIPTION

[0049] Reference will now be made in detail to exemplary embodiments of the present invention, examples of which are illustrated in the accompanying drawings. Whenever possible, the same reference numerals are used in the drawings and the description to refer to the same or like parts.

[0050] Figure 1 A schematic diagram of an image detection device 100 is shown according to an embodiment of the present invention. The image detection device 100 is used to detect one or more objects in an image. The image detection device 100 may include a processor 110 , a storage medium 120 , and a transceiver 130 .

[0051] The processor 110 is, for example, a central processing unit (CPU), or other programmable general-purpose or special-purpose micro control unit (MCU), microprocessor, digital signal processor (DSP), programmable controller, application specific integrated circuit (ASIC), graphics processing unit (GPU), image signal processor (ISP), image processing unit (IPU), arithmetic logic unit (ALU), complex programmable logic device (CPLD), field programmable gate array (FPGA), or other similar components or combinations thereof. The processor 110 may be coupled to the storage medium 120 and the transceiver 130, and access and execute multiple modules and various applications stored in the storage medium 120.

[0052] The storage medium 120 is, for example, any type of fixed or removable random access memory (RAM), read-only memory (ROM), flash memory, hard disk drive (HDD), solid state drive (SSD) or similar components or a combination of the above components, and is used to store multiple modules or various applications that can be executed by the processor 110. In this embodiment, the storage medium 120 can store multiple modules including a data collection module 121, an image processing module 122, and an image detection module 123, and their functions will be described later.

[0053] The transceiver 130 transmits and receives signals in a wireless or wired manner. The transceiver 130 may also perform operations such as low noise amplification, impedance matching, mixing, up or down frequency conversion, filtering, amplification, and the like.

[0054] The data collection module 121 can obtain the image 200 through the transceiver 130. For example, the transceiver 130 can be connected to a camera or a cloud server. The data collection module 121 can obtain the image 200 from the camera or the cloud server through the transceiver 130.

[0055] In one embodiment, due to the limitation of the hardware architecture of the image detection device 100, the image detection module 123 for detecting objects in an image only supports input images of a specific size. In order to enable the image detection module 123 to efficiently detect objects in the image 200, the image processing module 122 may perform image processing on the image 200 in advance to generate an input image whose size matches that of the image detection module 123.

[0056] Specifically, the image processing module 122 may adjust the size of the image 200 to generate an adjusted image 300, such as Figure 2 shown. Figure 2 According to an embodiment of the present invention, a schematic diagram of generating an adjusted image 300 is shown. Assuming that the size of the input image of the image detection module 123 is li×wi and the size of the image 200 is l0×w0, the image processing module 122 can scale the image 200 to generate an adjusted image 300 of size li×wi. In this embodiment, the adjusted image 300 corresponds to, for example, frame #t0 of the image 200. In other words, the image processing module 122 can scale the frame #t0 of the image 200 to generate the adjusted image 300.

[0057] The adjusted image 300 may include all information in the image 200, and is more suitable for detecting objects that are closer in the image 200. For example, assuming that the size li×wi of the adjusted image 300 is 320×224, the image processing module 122 can accurately detect objects within 6 meters in the image 200 based on the adjusted image 300.

[0058] On the other hand, in order to detect objects at a relatively far distance in the image 200 , the image processing module 122 may generate a plurality of segmented images according to the image 200 , wherein each of the plurality of segmented images may include partial information of the image 200 . Figure 3 According to an embodiment of the present invention, a schematic diagram of generating segmented images 410 and 420 is shown. If the image processing module 122 intends to generate two segmented images corresponding to the image 200 so as to use the two segmented images as input images of the image detection module 123, the image processing module 122 may first adjust the size of the image 200 to generate an adjusted image 400 (the size l1×w1 of the adjusted image 400 is, for example, equal to 2li×w1), and then segment the adjusted image 400 to generate segmented images 410 and 420 of sizes li×wi, as shown in FIG. Figure 3 As shown, where wi may be less than or equal to w1.

[0059] It is worth noting that if the image 200 can be directly segmented into the segmented images 410 and the segmented images 420 that match the size of the input image (for example: the size of the image 200 l0×w0 is equal to 2li×w1), then the image processing module 122 will not need to generate the adjusted image 400. The image processing module 122 can directly segment the image 200 to generate the segmented images 410 and the segmented images 420.

[0060] In this embodiment, the segmented image 410 corresponds to, for example, the frame #t1 of the image 200, and the segmented image 420 corresponds to, for example, the frame #t2 of the image 200, wherein the frame #t0, the frame #t1, and the frame #t2 are not the same. In other words, the image processing module 122 can generate the segmented image 410 according to the frame #t1 of the image 200, and can generate the segmented image 420 according to the frame #t2 of the image 200.

[0061] The segmented image 410 (or the segmented image 420) may include part of the information in the image 200, which is more suitable for detecting objects that are farther away in the image 200. For example, assuming that the size li×wi of the segmented image 410 is 320×224, the image processing module 122 can accurately detect objects within 11 meters in the image 200 based on the segmented image 410.

[0062] According to Figure 3In the embodiment shown, if the object is located at the junction of multiple segmented images (eg, segmented images 410 and 420), the object may not be successfully detected by the image detection module 123. Figure 4A and Figure 4B A schematic diagram showing generating a segmented image 410 according to another embodiment of the present invention is shown.

[0063] When the image processing module 122 generates two segmented images according to the image 200, the two segmented images need to include overlapping portions respectively to prevent objects located at the junction of the segmented images from being unable to be detected, wherein the overlapping portions can be adjusted according to the type of object that the user wants to detect. Specifically, the image processing module 122 can select a corresponding reference image (e.g., from a plurality of images pre-stored in the storage medium 120) according to the type of object that the user wants to detect, and decide how to segment the image 200 according to the reference image. Figure 5 A schematic diagram of a reference image 700 is shown according to an embodiment of the present invention. For example, assuming that the user wants to detect a person in the image 200, the image processing module 122 may select a reference image 700 corresponding to the person, wherein the size of the reference image 700 is lr×wr, and the reference image 700 may include at least one reference object 1. It is worth noting that the aspect ratio of the reference image 700 may be the same as the aspect ratio of the input image of the image detection module 123.

[0064] Please also refer to Figure 4A as well as Figure 5 , the image processing module 122 may adjust the size of the image 200 according to the size of the reference image 700 to generate an adjusted image 500, wherein the size of the adjusted image 500 may be the same as the size of the reference image 700. Next, the image processing module 122 may segment the adjusted image 500 according to the size of the reference object 1 in the reference image 700 to generate a segmented image 510 and a segmented image 520. In one embodiment, if the length of the reference object 1 in the reference image 700 is 1d, then two adjacent segmented images 510 and 520 of the adjusted image 500 must at least include an overlapping area 530 of a length of 1d, respectively. In this way, it can be ensured that the object corresponding to the reference object 1 in the adjusted image 500 can be completely included in at least one of the segmented image 510 and the segmented image 520.

[0065] After generating the segmented image 510 and the segmented image 520, the image processing module 122 may adjust the sizes of the segmented image 510 and the segmented image 520 to generate a plurality of input images for the image detection module 123. Figure 4BFor example, the image processing module 122 may resize the segmented image 510 to li×wi, thereby generating the segmented image 410. Similarly, the image processing module 122 may resize the segmented image 520 to li×wi, thereby generating the segmented image 420.

[0066] After the adjusted image 300 , the segmented image 410 , and the segmented image 420 are generated, the adjusted image 300 , the segmented image 410 , and the segmented image 420 may be input as a plurality of input images to the image detection module 123 , and the image detection module 123 may detect objects in the image 200 based on the plurality of input images.

[0067] Figure 6 According to an embodiment of the present invention, a schematic diagram of detecting an object in an image 200 according to candidate frames of different frames is shown. The image detection module 123 can identify an object in frame #t0 of the image 200 according to the adjusted image 300 corresponding to frame #t0 of the image 200. Specifically, in process P1, the image detection module 123 can use image recognition technology to detect at least one object in the adjusted image 300, thereby generating at least one candidate frame corresponding to the input image of frame #t0 (i.e., the adjusted image 300).

[0068] Next, in process P2, the image detection module 123 may determine at least one candidate frame for detecting an object in frame #t0 of the image 200 based on at least one candidate frame of the input image corresponding to frame #t0 (i.e., the adjusted image) and at least one candidate frame of the input image corresponding to the previous frame of frame #t0. Since frame #t0 in this embodiment is the first frame, the image detection module 123 may determine at least one candidate frame corresponding to frame #t0 of the image 200 based only on at least one candidate frame of the input image corresponding to frame #t0 (i.e., the adjusted image 300). For example, the at least one candidate frame for detecting an object in frame #t0 of the image 200 may include candidate frame 601 and candidate frame 602 corresponding to the adjusted image 300.

[0069] Then, in process P3, the image detection module 123 may detect the object in the frame #t0 of the image 200 according to the candidate frame 601 and the candidate frame 602. For example, the image detection module 123 may generate the target frame 600 according to the candidate frame 601 and the candidate frame 602 based on the non-maximum suppression (NMS) algorithm. Finally, the image detection module 123 may identify the object in the frame #t0 of the image 200 according to the target frame 600.

[0070] Similarly, the image detection module 123 may identify an object in the frame #t1 of the image 200 according to the segmented image 410 corresponding to the frame #t1 of the image 200. Specifically, in process P1, the image detection module 123 may detect at least one object in the segmented image 410 using image recognition technology, thereby generating at least one candidate frame corresponding to the input image of the frame #t1 (i.e., the segmented image 410).

[0071] Next, in process P2, the image detection module 123 may determine at least one candidate frame for detecting an object in frame #t1 of the image 200 according to at least one candidate frame corresponding to the input image of frame #t1 (i.e., the segmented image 410) and at least one candidate frame corresponding to the input image of frame #t0 (i.e., the adjusted image 300). For example, at least one candidate frame for detecting an object in frame #t1 of the image 200 may include candidate frames 601 and 602 corresponding to the adjusted image 300 and candidate frames 611, 612, and 613 corresponding to the segmented image 410.

[0072] Then, in process P3, the image detection module 123 may detect the object in the frame #t1 of the image 200 according to the candidate frame 601, the candidate frame 602, the candidate frame 611, the candidate frame 612, and the candidate frame 613. For example, the image detection module 123 may generate the target frame 610 according to the candidate frame 601, the candidate frame 602, and the candidate frame 611 based on the non-maximum suppression algorithm, and may generate the target frame 620 according to the candidate frame 612 and the candidate frame 613 based on the non-maximum suppression algorithm. Finally, the image detection module 123 may identify the object in the frame #t1 of the image 200 according to the target frame 610 and the target frame 620.

[0073] Similarly, the image detection module 123 may identify an object in the frame #t2 of the image 200 according to the segmented image 420 corresponding to the frame #t2 of the image 200. Specifically, in process P1, the image detection module 123 may detect at least one object in the segmented image 420 using image recognition technology, thereby generating at least one candidate frame corresponding to the input image of the frame #t2 (i.e., the segmented image 420).

[0074] Next, in process P2, the image detection module 123 may determine at least one candidate frame for detecting an object in frame #t2 of the image 200 based on at least one candidate frame corresponding to the input image of frame #t2 (i.e., segmented image 420) and at least one candidate frame corresponding to the input image of frame #t1 (i.e., segmented image 410). For example, at least one candidate frame for detecting an object in frame #t2 of the image 200 may include candidate frames 611, 612, and 613 corresponding to the segmented image 410 and candidate frames 621 and 622 corresponding to the segmented image 420.

[0075] Then, in process P3, the image detection module 123 may detect the object in the frame #t2 of the image 200 according to the candidate frame 611, the candidate frame 612, the candidate frame 613, the candidate frame 621, and the candidate frame 622. For example, the image detection module 123 may generate a target frame 640 according to the candidate frame 611 based on the non-maximum suppression algorithm, may generate a target frame 620 according to the candidate frame 612 and the candidate frame 613 based on the non-maximum suppression algorithm, and may generate a target frame 630 according to the candidate frame 621 and the candidate frame 622 based on the non-maximum suppression algorithm. Finally, the image detection module 123 may identify the object in the frame #t2 of the image 200 according to the target frame 620, the target frame 630, and the target frame 640.

[0076] Figure 7 According to an embodiment of the present invention, a schematic diagram of detecting an object in an image 200 according to a target frame and a candidate frame is shown. The image detection module 123 can identify an object in frame #t0 of the image 200 according to the adjusted image 300 corresponding to frame #t0 of the image 200. Specifically, in process P1, the image detection module 123 can use image recognition technology to detect at least one object in the adjusted image 300, thereby generating at least one candidate frame corresponding to the input image of frame #t0 (i.e., the adjusted image 300).

[0077] Next, in process P2, the image detection module 123 may determine at least one candidate frame for detecting an object in frame #t0 of the image 200 based on at least one candidate frame of the input image corresponding to frame #t0 (i.e., the adjusted image) and at least one target frame of the input image corresponding to the previous frame of frame #t0. Since frame #t0 in this embodiment is the first frame, the image detection module 123 may determine at least one candidate frame corresponding to frame #t0 of the image 200 based only on at least one candidate frame of the input image corresponding to frame #t0 (i.e., the adjusted image 300). For example, the at least one candidate frame for detecting an object in frame #t0 of the image 200 may include candidate frame 701 and candidate frame 702 corresponding to the adjusted image 300.

[0078] Then, in process P3, the image detection module 123 may detect the object in the frame #t0 of the image 200 according to the candidate frame 701 and the candidate frame 702. For example, the image detection module 123 may generate the target frame 700 according to the candidate frame 701 and the candidate frame 702 based on the non-maximum suppression algorithm. Finally, the image detection module 123 may identify the object in the frame #t0 of the image 200 according to the target frame 700.

[0079] Similarly, the image detection module 123 may identify an object in the frame #t1 of the image 200 according to the segmented image 410 corresponding to the frame #t1 of the image 200. Specifically, in process P1, the image detection module 123 may detect at least one object in the segmented image 410 using image recognition technology, thereby generating at least one candidate frame corresponding to the input image of the frame #t1 (i.e., the segmented image 410).

[0080] Next, in process P2, the image detection module 123 may determine at least one candidate frame for detecting an object in frame #t1 of the image 200 according to at least one candidate frame corresponding to the input image of frame #t1 (i.e., the segmented image 410) and at least one target frame corresponding to the input image of frame #t0 (i.e., the adjusted image 300). For example, at least one candidate frame for detecting an object in frame #t1 of the image 200 may include the target frame 700 corresponding to the adjusted image 300 and the candidate frames 711, 712, and 713 corresponding to the segmented image 410.

[0081] Then, in process P3, the image detection module 123 may detect the object in the frame #t1 of the image 200 according to the target frame 700, the candidate frame 711, the candidate frame 712, and the candidate frame 713. For example, the image detection module 123 may generate the target frame 710 according to the target frame 700 and the candidate frame 711 based on the non-maximum suppression algorithm, and may generate the target frame 720 according to the candidate frame 712 and the candidate frame 713 based on the non-maximum suppression algorithm. Finally, the image detection module 123 may identify the object in the frame #t1 of the image 200 according to the target frame 710 and the target frame 720.

[0082] Similarly, the image detection module 123 may identify an object in the frame #t2 of the image 200 according to the segmented image 420 corresponding to the frame #t2 of the image 200. Specifically, in process P1, the image detection module 123 may detect at least one object in the segmented image 420 using image recognition technology, thereby generating at least one candidate frame corresponding to the input image of the frame #t2 (i.e., the segmented image 420).

[0083] Next, in process P2, the image detection module 123 may determine at least one candidate frame for detecting an object in frame #t2 of the image 200 based on at least one candidate frame corresponding to the input image of frame #t2 (i.e., the segmented image 420) and at least one target frame corresponding to the input image of frame #t1 (i.e., the segmented image 410). For example, at least one candidate frame for detecting an object in frame #t2 of the image 200 may include target frame 710 and target frame 720 corresponding to the segmented image 410 and candidate frame 721 and candidate frame 722 corresponding to the segmented image 420.

[0084] Then, in process P3, the image detection module 123 may detect the object in the frame #t2 of the image 200 according to the target frame 710, the target frame 720, the candidate frame 721, and the candidate frame 722. For example, the image detection module 123 may generate the target frame 730 according to the candidate frame 721 and the candidate frame 722 based on the non-maximum suppression algorithm, may generate the target frame 740 according to the candidate frame 710 based on the non-maximum suppression algorithm, and may generate the target frame 750 according to the candidate frame 720 based on the non-maximum suppression algorithm. Finally, the image detection module 123 may identify the object in the frame #t2 of the image 200 according to the target frame 730, the target frame 740, and the target frame 750.

[0085] Figure 8 A schematic diagram of an image detection method is shown according to an embodiment of the present invention, wherein the image detection method can be performed as follows: Figure 1 The image detection device 100 shown in the figure is implemented. In step S801, an image is obtained, wherein the image includes an object. In step S802, a first size of the image is adjusted to generate an adjusted image. In step S803, a first segmented image and a second segmented image are generated according to the image. In step S804, an object in the image is detected according to a plurality of input images, wherein the plurality of input images include the first segmented image, the second segmented image, and the adjusted image.

[0086] In summary, the present invention can convert an image into multiple input images such as an adjusted image and a segmented image. The image detection device can use the adjusted image generated by enlarging the image to detect objects that are closer in the image. In addition, the image detection device can use the segmented image to detect objects that are farther away in the image. In any frame of the image, the image detection device will only detect one of the multiple input images. Therefore, the image detection device does not need to consume a lot of computing power, memory capacity, frequency or power, and can also achieve near real-time detection. Replacing the entire image with a segmented image as the input image of the image detection module will enable the image detection module to obtain more details of the image, and enable the image detection module to detect objects that are farther away in the image.

[0087] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit it. Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the aforementioned embodiments, or replace some or all of the technical features therein with equivalents. However, these modifications or replacements do not cause the essence of the corresponding technical solutions to deviate from the scope of the technical solutions of the embodiments of the present invention.

Claims

1. An image detection device, characterized in that: include: Transceiver; A storage medium for storing multiple modules; as well as A processor is coupled to the storage medium and the transceiver, and accesses and executes the multiple modules, wherein the multiple modules include: a data collection module, which acquires an image through the transceiver, wherein the image includes an object; an image processing module, adjusting a first size of the image to generate an adjusted image, and generating a first segmented image and a second segmented image according to the image; and An image detection module detects the object in the image based on a plurality of input images, wherein the plurality of input images include the first segmented image, the second segmented image, and the adjusted image, wherein the image processing module segments a first frame of the image to generate the first segmented image, and adjusts the first size of a third frame of the image to generate the adjusted image, wherein the first frame is different from the third frame. 2 . The image detection device according to claim 1 , wherein the image processing module segments a second frame of the image to generate the second segmented image, wherein the first frame is different from the second frame.

3. The image detection device according to claim 1, wherein the multiple input images correspond to different frames respectively, and wherein the image detection module detects the object in the first frame of the image based on at least one first candidate frame corresponding to the first frame in the multiple input images. 4 . The image detection device according to claim 3 , wherein the image detection module detects the object in the second frame of the image according to the at least one first candidate frame and at least one second candidate frame corresponding to the second frame of the plurality of input images.

5. The image detection device according to claim 3, wherein the image detection module determines a target frame based on the at least one first candidate frame, and detects the object in the second frame of the image based on the target frame and at least one second candidate frame corresponding to the second frame in the multiple input images. 6 . The image detection device according to claim 1 , wherein the image processing module adjusts the first size of the image to generate a second adjusted image, and segments the second adjusted image to generate the first segmented image and the second segmented image.

7. The image detection device according to claim 1, wherein the image processing module adjusts the first size of the image according to a reference image to generate a third adjusted image, segments the third adjusted image according to the size of a reference object corresponding to the object in the reference image to generate a third segmented image, and adjusts the second size of the third segmented image to generate the first segmented image, wherein the first aspect ratio of the reference image is the same as the second aspect ratio of any one of the multiple input images.

8. An image detection method, characterized in that: include: obtaining an image, wherein the image includes an object; Adjusting a first size of the image to produce an adjusted image includes: adjusting the first size of a third frame of the image to produce the adjusted image; Generating a first segmented image and a second segmented image according to the image, comprising: segmenting a first frame of the image to generate the first segmented image, wherein the first frame is different from the third frame; and The object in the image is detected according to a plurality of input images, wherein the plurality of input images include the first segmented image, the second segmented image, and the adjusted image.

9. The image detection method according to claim 8, wherein the step of generating the first segmented image and the second segmented image according to the image further comprises: A second frame of the image is segmented to generate the second segmented image, wherein the first frame is different from the second frame.

10. The image detection method according to claim 8, wherein the plurality of input images correspond to different frames respectively, and the step of detecting the object in the image according to the plurality of input images comprises: The object in the first frame of the image is detected according to at least one first candidate frame corresponding to the first frame of the plurality of input images.

11. The image detection method according to claim 10, wherein the step of detecting the object in the image according to the plurality of input images further comprises: The object in the second frame of the image is detected according to the at least one first candidate frame and at least one second candidate frame corresponding to a second frame of the plurality of input images.

12. The image detection method according to claim 10, wherein the step of detecting the object in the image according to the plurality of input images further comprises: A target frame is determined according to the at least one first candidate frame, and the object in the second frame of the image is detected according to the target frame and at least one second candidate frame corresponding to the second frame of the plurality of input images.

13. The image detection method according to claim 8, wherein the step of generating the first segmented image and the second segmented image according to the image further comprises: The first size of the image is adjusted to produce a second adjusted image, and the second adjusted image is segmented to produce the first segmented image and the second segmented image.

14. The image detection method according to claim 8, wherein the step of generating the first segmented image and the second segmented image according to the image comprises: adjusting the first size of the image according to a reference image to produce a third adjusted image; segmenting the third adjusted image according to a size of a reference object in the reference image corresponding to the object to generate a third segmented image; as well as A second size of the third segmented image is adjusted to generate the first segmented image, wherein a first aspect ratio of the reference image is the same as a second aspect ratio of any one of the plurality of input images.