CN112183291A - Method and system for detecting tiny object in image, storage medium and terminal - Google Patents
Method and system for detecting tiny object in image, storage medium and terminal Download PDFInfo
- Publication number
- CN112183291A CN112183291A CN202011004320.6A CN202011004320A CN112183291A CN 112183291 A CN112183291 A CN 112183291A CN 202011004320 A CN202011004320 A CN 202011004320A CN 112183291 A CN112183291 A CN 112183291A
- Authority
- CN
- China
- Prior art keywords
- feature map
- convolution
- pooling
- detecting
- nonlinear function
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V20/00—Scenes; Scene-specific elements
- G06V20/20—Scenes; Scene-specific elements in augmented reality scenes
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/04—Architecture, e.g. interconnection topology
- G06N3/045—Combinations of networks
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/06—Physical realisation, i.e. hardware implementation of neural networks, neurons or parts of neurons
- G06N3/063—Physical realisation, i.e. hardware implementation of neural networks, neurons or parts of neurons using electronic means
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V2201/00—Indexing scheme relating to image or video recognition or understanding
- G06V2201/07—Target detection
Abstract
The invention provides a method and a system for detecting a tiny object in an image, a storage medium and a terminal, wherein the method comprises the steps of obtaining an input image and performing convolution of at least four groups; the feature maps obtained by the last four groups of convolutions are respectively a first feature map, a second feature map, a third feature map and a fourth feature map; performing up-sampling on the second feature map for the first time to obtain a fifth feature map, performing up-sampling on the third feature map for the second time to obtain a sixth feature map, and performing up-sampling on the fourth feature map for the third time to obtain a seventh feature map; connecting the first characteristic diagram, the fifth characteristic diagram, the sixth characteristic diagram and the seventh characteristic diagram to obtain an eighth characteristic diagram; performing convolution on the eighth feature map to obtain a ninth feature map; performing convolution on the ninth feature map to obtain a tenth feature map; and detecting the extremely small object in the input image based on the tenth feature map. The method and the system for detecting the extremely small object in the image, the storage medium and the terminal realize the detection of the extremely small object in the image, and ensure the reliability and the effectiveness of the image detection.
Description
Technical Field
The present invention relates to the field of image processing technologies, and in particular, to a method and a system for detecting a very small object in an image, a storage medium, and a terminal.
Background
Object detection (object detection) is to find the position of an object in a given image and to mark the object type. In the prior art, a target detection algorithm is mainly based on a deep learning model and mainly includes the following two categories:
(1) one-stage detection algorithm does not need a candidate region (region pro-potential) stage, directly generates the class probability and the position coordinate value of an object, and can directly obtain a final detection result through single detection, so that the one-stage detection algorithm has a faster detection speed, and is more typical algorithms such as YOLO, SSD and Retina-Net.
(2) And a two-stage detection algorithm, namely generating a suggested target candidate Region from the input picture by using a corresponding Region Proposal algorithm, and sending all the candidate regions into a classifier for classification.
The main performance indicators of the target detection model are detection accuracy and speed, and for accuracy, target detection considers the positioning accuracy of an object, not just classification accuracy. In general, the two-stage algorithm has an advantage in accuracy, while the one-stage algorithm has an advantage in speed.
However, in either one-stage detection algorithm or two-stage detection algorithm, the detection of objects with a common size in a picture is only good, and the detection of very small objects with a size smaller than 20 × 20 pixels is not considered, so that the detection of very small objects is missed.
Disclosure of Invention
In view of the above-mentioned shortcomings of the prior art, an object of the present invention is to provide a method and a system for detecting a very small object in an image, a storage medium and a terminal, which are capable of detecting the very small object in the image based on a neural network capable of detecting the very small object, and ensuring reliability and effectiveness of image detection.
To achieve the above and other related objects, the present invention provides a method for detecting a very small object in an image, comprising the steps of: acquiring an input image; continuously performing at least four groups of convolution, pooling and nonlinear function activation operations on the input image; the feature maps obtained by the last four groups of convolution, pooling and nonlinear function activation operations are respectively a first feature map, a second feature map, a third feature map and a fourth feature map; performing up-sampling operation on the second feature map for the first time to obtain a fifth feature map, performing up-sampling operation on the third feature map for the second time to obtain a sixth feature map, and performing up-sampling operation on the fourth feature map for the third time to obtain a seventh feature map; connecting the first feature diagram, the fifth feature diagram, the sixth feature diagram and the seventh feature diagram to obtain an eighth feature diagram; performing convolution, pooling and nonlinear function activation operation on the eighth feature map for a first preset number of times to obtain a ninth feature map; performing convolution, pooling and nonlinear function activation operation on the ninth feature map for a second preset number of times to obtain a tenth feature map; detecting a very small object in the input image based on the tenth feature map.
In an embodiment of the present invention, the method further includes preprocessing the input image to perform six convolution, pooling and nonlinear function activation operations based on the preprocessed input image.
In an embodiment of the present invention, the last four sets of convolution, pooling and nonlinear function activation operations correspond to two, eight and four consecutive convolution, pooling and nonlinear function activation operations, respectively.
In an embodiment of the present invention, the first time, the second time and the third time are respectively one time, two times and three times; the first preset number is five, and the second preset number is one.
Correspondingly, the invention provides a system for detecting a very small object in an image, which comprises an acquisition module, a first convolution module, an up-sampling module, a connection module, a second convolution module, a third convolution module and a detection module, wherein the acquisition module is used for acquiring a first convolution signal;
the acquisition module is used for acquiring an input image;
the first convolution module is used for continuously performing at least four groups of convolution, pooling and nonlinear function activation operations on the input image; the feature maps obtained by the last four groups of convolution, pooling and nonlinear function activation operations are respectively a first feature map, a second feature map, a third feature map and a fourth feature map;
the up-sampling module is used for performing up-sampling operation on the second feature map for a first time to obtain a fifth feature map, performing up-sampling operation on the third feature map for a second time to obtain a sixth feature map, and performing up-sampling operation on the fourth feature map for a third time to obtain a seventh feature map;
the connection module is configured to connect the first feature map, the fifth feature map, the sixth feature map, and the seventh feature map to obtain an eighth feature map;
the second convolution module is used for performing convolution, pooling and nonlinear function activation operation on the eighth feature map for a first preset number of times to obtain a ninth feature map;
the third convolution module is used for performing convolution, pooling and nonlinear function activation operation on the ninth feature map for a second preset number of times to obtain a tenth feature map;
the detection module is used for detecting a tiny object in the input image based on the tenth feature map.
In an embodiment of the present invention, the method further includes preprocessing the input image to perform six convolution, pooling and nonlinear function activation operations based on the preprocessed input image.
In an embodiment of the present invention, the last four sets of convolution, pooling and nonlinear function activation operations correspond to two, eight and four consecutive convolution, pooling and nonlinear function activation operations, respectively.
In an embodiment of the present invention, the first time, the second time and the third time are respectively one time, two times and three times; the first preset number is five, and the second preset number is one.
The present invention provides a storage medium having stored thereon a computer program which, when executed by a processor, implements the above-described method of detecting a small object in an image.
Finally, the present invention provides a terminal comprising: a processor and a memory;
the memory is used for storing a computer program;
the processor is used for executing the computer program stored in the memory so as to enable the terminal to execute the method for detecting the extremely small object in the image.
As described above, the method and system for detecting a very small object in an image, the storage medium, and the terminal according to the present invention have the following advantageous effects:
(1) the method comprises the steps of realizing the detection of the tiny object in an image based on a neural network capable of detecting the tiny object;
(2) the missing detection of the extremely small object in the image detection is avoided, and the reliability and effectiveness of the image detection are ensured.
Drawings
FIG. 1 is a flow chart illustrating a method for detecting a very small object in an image according to an embodiment of the present invention;
FIG. 2 is a block diagram of a method for detecting a very small object in an image according to an embodiment of the present invention;
FIG. 3 is a schematic diagram illustrating an embodiment of a system for detecting small objects in an image according to the present invention;
fig. 4 is a schematic structural diagram of a terminal according to an embodiment of the invention.
Description of the element reference numerals
31 acquisition module
32 first volume module
33 upsampling module
34 connection module
35 second convolution module
36 third convolution module
37 detection module
41 processor
42 memory
Detailed Description
The embodiments of the present invention are described below with reference to specific embodiments, and other advantages and effects of the present invention will be easily understood by those skilled in the art from the disclosure of the present specification. The invention is capable of other and different embodiments and of being practiced or of being carried out in various ways, and its several details are capable of modification in various respects, all without departing from the spirit and scope of the present invention.
It should be noted that the drawings provided in the present embodiment are only for illustrating the basic idea of the present invention, and the components related to the present invention are only shown in the drawings rather than drawn according to the number, shape and size of the components in actual implementation, and the type, quantity and proportion of the components in actual implementation may be changed freely, and the layout of the components may be more complicated.
The method and the system for detecting the extremely small object in the image, the storage medium and the terminal realize the detection of the extremely small object in the image through the neural network for detecting the extremely small object, effectively avoid the omission of the extremely small object, and further ensure the reliability and the effectiveness of the image detection.
As shown in fig. 1, in an embodiment, the method for detecting a very small object in an image of the present invention includes the following steps:
step S1, an input image is acquired.
Specifically, the input image contains very small objects. Preferably, the very small objects are objects smaller than 20 x 20 pixels.
Step S2, continuously performing at least four groups of operations of convolution, pooling and nonlinear function activation on the input image; and the feature maps obtained by the last four groups of convolution, pooling and nonlinear function activation operations are respectively a first feature map, a second feature map, a third feature map and a fourth feature map.
Specifically, as shown in fig. 2, taking an input image of 416 x 3 pixels as an example,
first, an input image of 416 × 3 pixels is subjected to a convolution operation, a pooling operation, and a nonlinear function activation operation using a convolution kernel of 32 × 3, and a feature map (feature map) of 416 × 41 × 32 pixels is acquired.
Next, a convolution kernel of 64 × 3 size was used to perform a convolution operation, a pooling operation, and a nonlinear function activation operation on the feature map of 416 × 41 × 32 pixels, and thereby 208 × 64 pixels were acquired.
Then, using a convolution kernel of 128 × 3 size, the convolution operation, the pooling operation, and the nonlinear function activation operation are performed on the feature map of 208 × 64 pixels twice in succession, and the feature map a of 104 × 128 pixels is acquired.
Then, using a convolution kernel of 256 × 3, the convolution operation, pooling operation, and nonlinear function activation operation are performed on the feature map a of 104 × 128 pixels, eight times in succession, and the feature map B of 52 × 256 pixels is acquired.
Then, the convolution kernel of 512 × 3 size is used to perform the convolution operation, pooling operation, and nonlinear function activation operation on the feature map B of 52 × 256 pixels eight times in succession, and the feature map C of 26 × 512 pixels is acquired.
Finally, convolution kernels of 1024 × 3 are used to perform convolution operation, pooling operation, and nonlinear function activation operation on the feature map C of 26 × 512 pixels four times in succession, and the feature map D of 13 × 1024 pixels is acquired.
Step S3, performing a first-time upsampling operation on the second feature map to obtain a fifth feature map, performing a second-time upsampling operation on the third feature map to obtain a sixth feature map, and performing a third-time upsampling operation on the fourth feature map to obtain a seventh feature map.
Specifically, performing an upsampling operation on the feature map B once to obtain a feature map E of 104 × 256 pixels; performing two upsampling operations on the feature map C to obtain a feature map F of 104 × 512 pixels; and performing up-sampling operation on the feature map D three times to obtain a feature map G with 104 × 1024 pixels.
And step S4, connecting the first feature map, the fifth feature map, the sixth feature map and the seventh feature map to obtain an eighth feature map.
Specifically, a connection (concat) operation is performed on feature map a, feature map E, feature map F, and feature map G, obtaining feature map H of 104 × 1920 pixels.
And step S5, performing convolution, pooling and nonlinear function activation operation for a first preset number of times on the eighth feature map to obtain a ninth feature map.
Specifically, a convolution kernel of 128 × 3 size is used to perform a convolution operation, a pooling operation, and a nonlinear function activation operation on the feature map H of 104 × 1920 pixels five times in succession, and the feature map J of 104 × 128 pixels is acquired.
And step S6, performing convolution, pooling and nonlinear function activation operation for a second preset time on the ninth feature map to obtain a tenth feature map.
Specifically, a convolution kernel of 75 × 3 size is used to perform a convolution operation, a pooling operation, and activation of a nonlinear function on the feature map J of 104 × 128 pixels, and thereby obtain the feature map K of 104 × 75 pixels.
Step S7, detecting a very small object in the input image based on the tenth feature map.
Specifically, a feature map K of 104 × 75 pixels is used to detect a very small object in the input image.
In an embodiment of the present invention, the method for detecting a very small object in an image further includes preprocessing the input image, and performing convolution, pooling, and nonlinear function activation operations six times based on the preprocessed input image, so as to ensure the accuracy of the detection of the very small object. Preferably, the preprocessing operation includes one or a combination of denoising, contrast enhancement.
As shown in fig. 3, in an embodiment, the system for detecting a very small object in an image according to the present invention includes an acquisition module 31, a first convolution module 32, an upsampling module 33, a connection module 34, a second convolution module 35, a third convolution module 36, and a detection module 37.
The acquiring module 31 is used for acquiring an input image.
The first convolution module 32 is connected to the obtaining module 31, and is configured to perform at least four sets of convolution, pooling and nonlinear function activation operations on the input image; and the feature maps obtained by the last four groups of convolution, pooling and nonlinear function activation operations are respectively a first feature map, a second feature map, a third feature map and a fourth feature map.
The upsampling module 33 is connected to the first convolution module 32, and configured to perform upsampling on the second feature map for a first time to obtain a fifth feature map, perform upsampling on the third feature map for a second time to obtain a sixth feature map, and perform upsampling on the fourth feature map for a third time to obtain a seventh feature map.
The connection module 34 is connected to the first convolution module 32 and the upsampling module 33, and is configured to connect the first feature map, the fifth feature map, the sixth feature map, and the seventh feature map to obtain an eighth feature map.
The second convolution module 35 is connected to the connection module 34, and configured to perform convolution, pooling and nonlinear function activation operations for a first preset number of times on the eighth feature map to obtain a ninth feature map.
The third convolution module 36 is connected to the second convolution module 35, and configured to perform convolution, pooling and nonlinear function activation operations for a second preset number of times on the ninth feature map, so as to obtain a tenth feature map.
The detection module 37 is connected to the third convolution module 36, and is configured to detect a very small object in the input image based on the tenth feature map.
The structures and principles of the obtaining module 31, the first convolution module 32, the upsampling module 33, the connection module 34, the second convolution module 35, the third convolution module 36 and the detection module 37 correspond to the steps in the method for detecting a very small object in an image one to one, and therefore are not described herein again.
It should be noted that the division of the modules of the above apparatus is only a logical division, and the actual implementation may be wholly or partially integrated into one physical entity, or may be physically separated. And these modules can be realized in the form of software called by processing element; or may be implemented entirely in hardware; and part of the modules can be realized in the form of calling software by the processing element, and part of the modules can be realized in the form of hardware. For example, the x module may be a processing element that is set up separately, or may be implemented by being integrated in a chip of the apparatus, or may be stored in a memory of the apparatus in the form of program code, and the function of the x module may be called and executed by a processing element of the apparatus. Other modules are implemented similarly. In addition, all or part of the modules can be integrated together or can be independently realized. The processing element described herein may be an integrated circuit having signal processing capabilities. In implementation, each step of the above method or each module above may be implemented by an integrated logic circuit of hardware in a processor element or an instruction in the form of software.
For example, the above modules may be one or more integrated circuits configured to implement the above methods, such as: one or more Application Specific Integrated Circuits (ASICs), or one or more microprocessors (DSPs), or one or more Field Programmable Gate Arrays (FPGAs), among others. For another example, when one of the above modules is implemented in the form of a Processing element scheduler code, the Processing element may be a general-purpose processor, such as a Central Processing Unit (CPU) or other processor capable of calling program code. For another example, these modules may be integrated together and implemented in the form of a system-on-a-chip (SOC).
The storage medium of the present invention stores thereon a computer program, which is characterized by implementing the above-described method for detecting a small object in an image when the program is executed by a processor. The storage medium includes: various media that can store program codes, such as ROM, RAM, magnetic disk, U-disk, memory card, or optical disk.
As shown in fig. 4, in an embodiment, the terminal of the present invention includes: a processor 41 and a memory 42.
The memory 42 is used for storing computer programs.
The memory 42 includes: various media that can store program codes, such as ROM, RAM, magnetic disk, U-disk, memory card, or optical disk.
The processor 41 is connected to the memory 42, and is configured to execute the computer program stored in the memory 42, so that the terminal executes the above-mentioned method for detecting a very small object in an image.
Preferably, the Processor 41 may be a general-purpose Processor, including a Central Processing Unit (CPU), a Network Processor (NP), and the like; the Integrated Circuit may also be a Digital Signal Processor (DSP), an Application Specific Integrated Circuit (ASIC), a Field Programmable Gate Array (FPGA) or other Programmable logic device, discrete Gate or transistor logic device, or discrete hardware components.
In summary, the method and system for detecting a very small object in an image, the storage medium, and the terminal of the present invention realize the detection of the very small object in the image based on the neural network capable of detecting the very small object; the missing detection of the extremely small object in the image detection is avoided, and the reliability and effectiveness of the image detection are ensured. Therefore, the invention effectively overcomes various defects in the prior art and has high industrial utilization value.
The foregoing embodiments are merely illustrative of the principles and utilities of the present invention and are not intended to limit the invention. Any person skilled in the art can modify or change the above-mentioned embodiments without departing from the spirit and scope of the present invention. Accordingly, it is intended that all equivalent modifications or changes which can be made by those skilled in the art without departing from the spirit and technical spirit of the present invention be covered by the claims of the present invention.
Claims (10)
1. A method for detecting a tiny object in an image is characterized by comprising the following steps: the method comprises the following steps:
acquiring an input image;
continuously performing at least four groups of convolution, pooling and nonlinear function activation operations on the input image; the feature maps obtained by the last four groups of convolution, pooling and nonlinear function activation operations are respectively a first feature map, a second feature map, a third feature map and a fourth feature map;
performing up-sampling operation on the second feature map for the first time to obtain a fifth feature map, performing up-sampling operation on the third feature map for the second time to obtain a sixth feature map, and performing up-sampling operation on the fourth feature map for the third time to obtain a seventh feature map;
connecting the first feature diagram, the fifth feature diagram, the sixth feature diagram and the seventh feature diagram to obtain an eighth feature diagram;
performing convolution, pooling and nonlinear function activation operation on the eighth feature map for a first preset number of times to obtain a ninth feature map;
performing convolution, pooling and nonlinear function activation operation on the ninth feature map for a second preset number of times to obtain a tenth feature map;
detecting a very small object in the input image based on the tenth feature map.
2. The method for detecting very small objects in an image according to claim 1, characterized in that: the method further comprises preprocessing the input image to perform six convolution, pooling and nonlinear function activation operations based on the preprocessed input image.
3. The method for detecting very small objects in an image according to claim 1, characterized in that: the last four groups of convolution, pooling and nonlinear function activation operations correspond to two, eight and four consecutive convolution, pooling and nonlinear function activation operations, respectively.
4. The method for detecting very small objects in an image according to claim 1, characterized in that: the first time, the second time and the third time are respectively one time, two times and three times; the first preset number is five, and the second preset number is one.
5. A system for detecting a small object in an image, comprising: the device comprises an acquisition module, a first convolution module, an up-sampling module, a connection module, a second convolution module, a third convolution module and a detection module;
the acquisition module is used for acquiring an input image;
the first convolution module is used for continuously performing at least four groups of convolution, pooling and nonlinear function activation operations on the input image; the feature maps obtained by the last four groups of convolution, pooling and nonlinear function activation operations are respectively a first feature map, a second feature map, a third feature map and a fourth feature map;
the up-sampling module is used for performing up-sampling operation on the second feature map for a first time to obtain a fifth feature map, performing up-sampling operation on the third feature map for a second time to obtain a sixth feature map, and performing up-sampling operation on the fourth feature map for a third time to obtain a seventh feature map;
the connection module is configured to connect the first feature map, the fifth feature map, the sixth feature map, and the seventh feature map to obtain an eighth feature map;
the second convolution module is used for performing convolution, pooling and nonlinear function activation operation on the eighth feature map for a first preset number of times to obtain a ninth feature map;
the third convolution module is used for performing convolution, pooling and nonlinear function activation operation on the ninth feature map for a second preset number of times to obtain a tenth feature map;
the detection module is used for detecting a tiny object in the input image based on the tenth feature map.
6. The system for detecting small objects in an image according to claim 5, wherein: the method further comprises preprocessing the input image to perform six convolution, pooling and nonlinear function activation operations based on the preprocessed input image.
7. The system for detecting small objects in an image according to claim 5, wherein: the last four groups of convolution, pooling and nonlinear function activation operations correspond to two, eight and four consecutive convolution, pooling and nonlinear function activation operations, respectively.
8. The system for detecting small objects in an image according to claim 5, wherein: the first time, the second time and the third time are respectively one time, two times and three times; the first preset number is five, and the second preset number is one.
9. A storage medium on which a computer program is stored, characterized in that the program, when executed by a processor, implements the method for detecting a very small object in an image according to any one of claims 1 to 4.
10. A terminal, comprising: a processor and a memory;
the memory is used for storing a computer program;
the processor is used for executing the computer program stored in the memory to enable the terminal to execute the method for detecting the tiny object in the image according to any one of claims 1 to 4.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202011004320.6A CN112183291A (en) | 2020-09-22 | 2020-09-22 | Method and system for detecting tiny object in image, storage medium and terminal |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202011004320.6A CN112183291A (en) | 2020-09-22 | 2020-09-22 | Method and system for detecting tiny object in image, storage medium and terminal |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN112183291A true CN112183291A (en) | 2021-01-05 |
Family
ID=73955857
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202011004320.6A Pending CN112183291A (en) | 2020-09-22 | 2020-09-22 | Method and system for detecting tiny object in image, storage medium and terminal |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN112183291A (en) |
Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110287849A (en) * | 2019-06-20 | 2019-09-27 | 北京工业大学 | A kind of lightweight depth network image object detection method suitable for raspberry pie |
| CN111210417A (en) * | 2020-01-07 | 2020-05-29 | 创新奇智(北京)科技有限公司 | Cloth defect detection method based on convolutional neural network |
| CN111340049A (en) * | 2020-03-06 | 2020-06-26 | 清华大学 | Image processing method and device based on wide-area dynamic convolution |
| CN111553867A (en) * | 2020-05-15 | 2020-08-18 | 润联软件系统(深圳)有限公司 | Image deblurring method and device, computer equipment and storage medium |
-
2020
- 2020-09-22 CN CN202011004320.6A patent/CN112183291A/en active Pending
Patent Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110287849A (en) * | 2019-06-20 | 2019-09-27 | 北京工业大学 | A kind of lightweight depth network image object detection method suitable for raspberry pie |
| CN111210417A (en) * | 2020-01-07 | 2020-05-29 | 创新奇智(北京)科技有限公司 | Cloth defect detection method based on convolutional neural network |
| CN111340049A (en) * | 2020-03-06 | 2020-06-26 | 清华大学 | Image processing method and device based on wide-area dynamic convolution |
| CN111553867A (en) * | 2020-05-15 | 2020-08-18 | 润联软件系统(深圳)有限公司 | Image deblurring method and device, computer equipment and storage medium |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN109522874B (en) | Human body action recognition method and device, terminal equipment and storage medium | |
| CN109784372B (en) | Target classification method based on convolutional neural network | |
| CN112990228B (en) | Image feature matching method, related device, equipment and storage medium | |
| CN112085056B (en) | Target detection model generation method, device, equipment and storage medium | |
| CN111476719A (en) | Image processing method, image processing device, computer equipment and storage medium | |
| WO2022111609A1 (en) | Grid encoding method and computer system | |
| CN111860398A (en) | Remote sensing image target detection method and system and terminal equipment | |
| JP2021528726A (en) | Point cloud data processing methods, devices, electronic devices and storage media | |
| CN112435193A (en) | Method and device for denoising point cloud data, storage medium and electronic equipment | |
| CN114519868A (en) | Real-time bone key point identification method and system based on coordinate system regression | |
| CN110427506B (en) | Spatial data boundary processing method and device, computer equipment and storage medium | |
| CN110377342B (en) | Video memory processing method and device based on convolutional neural network and storage medium | |
| CN115331109A (en) | Remote sensing image target detection method based on rotation equal-variation convolution channel attention enhancement and multi-scale feature fusion | |
| CN110738317A (en) | FPGA-based deformable convolution network operation method, device and system | |
| US11704546B2 (en) | Operation processing apparatus that calculates addresses of feature planes in layers of a neutral network and operation processing method | |
| CN111967478B (en) | Feature map reconstruction method, system, storage medium and terminal based on weight overturn | |
| CN111639523B (en) | Target detection method, device, computer equipment and storage medium | |
| CN112183291A (en) | Method and system for detecting tiny object in image, storage medium and terminal | |
| WO2023060874A1 (en) | Picture classification and object detection synchronous processing method and system, storage medium, and terminal | |
| CN111507430A (en) | Feature coding method, device, equipment and medium based on matrix multiplication | |
| CN114757822B (en) | Binocular-based human body three-dimensional key point detection method and system | |
| CN108629219B (en) | Method and device for identifying one-dimensional code | |
| CN116051532A (en) | Deep learning-based industrial part defect detection method and system and electronic equipment | |
| CN115630663A (en) | Two-dimensional code identification method and device and electronic equipment | |
| CN111429388A (en) | Image processing method and device and terminal equipment |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination |