CN110992305A - Package counting method and system based on deep learning and multi-target tracking technology - Google Patents
Package counting method and system based on deep learning and multi-target tracking technology Download PDFInfo
- Publication number
- CN110992305A CN110992305A CN201911049870.7A CN201911049870A CN110992305A CN 110992305 A CN110992305 A CN 110992305A CN 201911049870 A CN201911049870 A CN 201911049870A CN 110992305 A CN110992305 A CN 110992305A
- Authority
- CN
- China
- Prior art keywords
- package
- parcel
- logistics
- target tracking
- counting
- 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
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/0002—Inspection of images, e.g. flaw detection
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/20—Analysis of motion
- G06T7/246—Analysis of motion using feature-based methods, e.g. the tracking of corners or segments
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/10—Image acquisition modality
- G06T2207/10016—Video; Image sequence
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/10—Image acquisition modality
- G06T2207/10024—Color image
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/20—Special algorithmic details
- G06T2207/20081—Training; Learning
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/20—Special algorithmic details
- G06T2207/20084—Artificial neural networks [ANN]
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/30—Subject of image; Context of image processing
- G06T2207/30242—Counting objects in image
Landscapes
- Engineering & Computer Science (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- Quality & Reliability (AREA)
- Multimedia (AREA)
- Image Analysis (AREA)
Abstract
The invention discloses a method and a system for counting parcels based on deep learning and multi-target tracking technology, which comprises the steps of firstly, acquiring parcel data of logistics and preprocessing the parcel data; training a preset package detection model based on deep learning through a training set, and testing and adjusting parameters of the trained package detection model through a verification set; obtaining a final package detection model; acquiring a real-time logistics package video, sampling key frames of the real-time logistics package video, and inputting a final package detection model for detection so as to obtain package position information of each key frame; and counting the packages by adopting a multi-target tracking algorithm according to the package position information of the key frames. The logistics parcel counting method based on the modern computer vision technology and the multi-target tracking counting is used for automatically counting logistics parcels, and by mining the parcel vision and position information of logistics, even if the parcel pictures acquired in real time contain densely and irregularly placed parcels, the parcels can be accurately counted.
Description
Technical Field
The invention relates to the technical field of parcel counting, in particular to a parcel counting method and system based on deep learning and multi-target tracking technology.
Background
In recent years, with the rapid development of electronic commerce and various shopping festival promotion activities, the order volume of the express logistics industry is increased like tsunami, and a plurality of challenges including package statistics, warehouse entry and exit, core and the like are brought to the traditional express industry. At present, the express logistics industry generally adopts a manual piece counting mode, but the manual piece counting mode has high strength and low efficiency; the sensor technology is dependent on distance measurement when the parcel quantity is large or dense, missed detection is often caused, and counting deviation of parcels is caused.
Disclosure of Invention
The invention provides a parcel counting method and system based on deep learning and multi-target tracking technology, aiming at solving the problems that counting is easy to deviate and counting efficiency is low in the existing logistics parcel counting mode.
In order to achieve the above purpose, the technical means adopted is as follows:
the parcel counting method based on the deep learning and multi-target tracking technology comprises the following steps:
s1, acquiring package data of logistics and preprocessing the package data; the logistics package data are divided into a training set and a verification set;
s2, training a preset deep learning-based package detection model through the training set, and testing and adjusting parameters of the trained package detection model through the verification set; obtaining a final package detection model;
s3, acquiring a real-time logistics package video, sampling key frames of the real-time logistics package video, and inputting the final package detection model for detection so as to obtain package position information of each key frame;
and S4, counting the packages by adopting a multi-target tracking algorithm according to the package position information of the key frames.
In the scheme, the logistics packages are automatically counted based on the modern computer vision technology and the multi-target tracking counting, and the packages can be accurately counted even if the packages which are densely and irregularly placed are contained in the package pictures which are acquired in real time by mining the package vision and position information of logistics.
Preferably, the parcel data of the logistics described in step S1 is obtained by arranging a camera device on the parcel conveyor belt, and acquiring continuous parcel pictures according to a preset frequency. In the preferred embodiment, the camera device may adopt an industrial camera, and take pictures so that the front and rear frames are continuous in content, and obtain the parcels at each position in the camera view.
Preferably, the package data obtained in step S1 includes 10000 package pictures, wherein 8000 package pictures are used as a training set, and 2000 package pictures are used as a verification set.
Preferably, the preprocessing in step S1 specifically includes: and marking the parcel boxes of the parcel pictures so as to obtain the position information of each parcel in each parcel picture, namely marking the coordinates of the upper left corner (x1, y1) and the coordinates of the lower right corner (x2, y2) of each parcel.
Preferably, the following steps are further included between step S1 and step S2:
performing data enhancement processing including horizontal flipping, rotation and noise addition on the training set;
and inputting the training set into a preset generation countermeasure network for sample expansion. In the preferred scheme, the samples of the training set are expanded, so that a detection model with higher robustness and generalization capability is obtained in subsequent training.
Preferably, the logistics package video is acquired in real time by arranging a camera device on the package conveyor belt.
Preferably, the specific step of step S4 is:
obtaining a parcel corresponding to the parcel position information according to the parcel position information of the key frame, respectively assigning a unique id to each parcel detected before the current frame, and finding and matching each id and the parcel detected by the current frame by adopting a linear assignment algorithm:
for the successfully matched id, updating the current package position of the id to be the package position detected by the current frame;
for id which is not matched successfully, the parcel corresponding to the id enters an undetermined area and the position of the parcel is updated by adopting an LSTM algorithm or a Kalman filtering algorithm;
for the id which is not successfully matched in the multi-frame, if the tracked times exceed a preset threshold value, ending the id and increasing the number of packages by 1; and if the tracked times do not exceed the preset threshold value, determining that the detection is invalid, and not increasing the count.
According to the package counting method based on the deep learning and multi-target tracking technology, the invention also provides a package counting system based on the deep learning and multi-target tracking technology, which comprises the following steps:
the data acquisition and pretreatment module is used for acquiring and pretreating the package data of the logistics; the logistics package data are divided into a training set and a verification set;
the package detection model training module is used for training a preset package detection model based on deep learning through the training set and testing and adjusting parameters of the trained package detection model through the verification set; obtaining a final package detection model;
the package detection module is used for acquiring a real-time logistics package video, sampling key frames of the real-time logistics package video, inputting the final package detection model for detection, and obtaining package position information of each key frame;
and the multi-target tracking parcel counting module is used for counting parcels by adopting a multi-target tracking algorithm according to the parcel position information of the key frame.
Compared with the prior art, the technical scheme of the invention has the beneficial effects that:
the logistics parcel counting method based on the modern computer vision technology and the multi-target tracking counting is used for automatically counting logistics parcels, and by mining the parcel vision and position information of logistics, even if the parcel pictures acquired in real time contain densely and irregularly placed parcels, the parcels can be accurately counted. The invention solves the problems that the counting is easy to deviate and the counting efficiency is low in the conventional logistics parcel counting mode based on manpower and hardware circuits.
Drawings
FIG. 1 is a general flow diagram of the process of the present invention.
FIG. 2 is a block diagram of the system of the present invention.
Detailed Description
The drawings are for illustrative purposes only and are not to be construed as limiting the patent;
for the purpose of better illustrating the embodiments, certain features of the drawings may be omitted, enlarged or reduced, and do not represent the size of an actual product;
it will be understood by those skilled in the art that certain well-known structures in the drawings and descriptions thereof may be omitted.
The technical solution of the present invention is further described below with reference to the accompanying drawings and examples.
Example 1
The parcel counting method based on the deep learning and multi-target tracking technology, as shown in fig. 1, includes the following steps:
s1, acquiring package data of logistics and preprocessing the package data; the logistics package data are divided into a training set and a verification set;
in the step, the industrial camera is arranged on the parcel data of the logistics on the parcel conveyor belt, continuous parcel pictures are obtained according to the preset frequency, all the logistics parcel data in a period of time are obtained, and in order to avoid missing logistics parcels, the front frame and the rear frame are continuous in content in the shooting process, so that parcels in all positions in the visual field of the camera can be obtained, the data distribution is avoided, and the model can be made to pay attention to all the positions of the pictures in the subsequent model training;
obtaining 10000 package pictures in total, wherein 8000 package pictures are used as a training set, and 2000 package pictures are used as a verification set; the parcel pictures comprise various parcel distribution pictures, namely dense parcel distribution, sparse distribution, edge distribution and the like, and various parcel shapes and the like;
marking the parcel pictures with parcel bounding boxes so as to obtain the position information of each parcel in each parcel picture, namely marking the upper left corner coordinates (x1, y1) and the lower right corner coordinates (x2, y2) of each parcel;
performing data enhancement processing including horizontal flipping, rotation and noise addition on the training set;
inputting the training set into a preset generation countermeasure network for sample expansion;
s2, training a preset deep learning-based package detection model through the training set, and testing and adjusting parameters of the trained package detection model through the verification set; obtaining a final package detection model; the package detection model adopts a model known in the deep learning field: a yolo series target detection model; in this embodiment 1, the picture of the input package detection model is a color picture with a size 416 × 3, the size of the trained batch is 64, a random gradient descent algorithm with momentum is adopted, the learning rate is 0.001, the momentum is 0.9, the learning rate attenuation factor is 0.0006, the maximum iteration number is 80000 times, and the package detection model is trained on the gpu server;
s3, acquiring a real-time logistics package video through a camera device arranged on a package conveyor belt, sampling key frames of the real-time logistics package video, and inputting the final package detection model for detection to obtain package position information of each key frame;
s4, counting the packages by adopting a multi-target tracking algorithm according to the package position information of the key frames; the method specifically comprises the following steps: obtaining a parcel corresponding to the parcel position information according to the parcel position information of the key frame, respectively assigning a unique id to each parcel detected before the current frame, and finding and matching each id and the parcel detected by the current frame by adopting a linear assignment algorithm:
for the successfully matched id, updating the current package position of the id to be the package position detected by the current frame;
for id which is not matched successfully, the parcel corresponding to the id enters an undetermined area and the position of the parcel is updated by adopting an LSTM algorithm or a Kalman filtering algorithm;
for the id which is not successfully matched in the multi-frame, if the tracked times exceed a preset threshold value, ending the id and increasing the number of packages by 1; and if the tracked times do not exceed the preset threshold value, determining that the detection is invalid, and not increasing the count.
Continuously allocating id to the just-entered parcel, repeating the above processes, realizing that the parcels in some frames are identified and found again under the condition of missed detection, and reducing the missed detection; meanwhile, because the tracking method is used in the embodiment, the position of the parcel can be tracked, so that the error detection which is occasionally generated, such as False Positive which suddenly appears in a certain visual field, can be eliminated, and the accuracy of parcel counting is further improved.
It should be noted that, in step S4, the existing modeling method of the motion model in the multi-target tracking field is adopted, and the visual information of the package is not used, because the model can solve the problem well for this scenario.
Example 2
The embodiment 2 is a package counting system based on deep learning and multi-target tracking technology, which is proposed by the package counting method based on deep learning and multi-target tracking technology of the embodiment 1, and as shown in fig. 2, the system includes:
the data acquisition and pretreatment module 1 is used for acquiring and pretreating package data of logistics; the logistics package data are divided into a training set and a verification set;
the package detection model training module 2 is used for training a preset package detection model based on deep learning through the training set and testing and adjusting parameters of the trained package detection model through the verification set; obtaining a final package detection model;
the parcel detection module 3 is used for acquiring a real-time logistics parcel video, sampling key frames of the real-time logistics parcel video, inputting the final parcel detection model for detection, and obtaining parcel position information of each key frame;
and the multi-target tracking parcel counting module 4 is used for counting parcels by adopting a multi-target tracking algorithm according to the parcel position information of the key frame.
The parcel technical system of this embodiment 2 can deploy to off-line logistics warehouse cooperation and use to the parcel figure on the accurate detection logistics conveyer belt can obtain the parcel total number through the logistics conveyer belt in a period of time through statistics.
The terms describing positional relationships in the drawings are for illustrative purposes only and are not to be construed as limiting the patent;
it should be understood that the above-described embodiments of the present invention are merely examples for clearly illustrating the present invention, and are not intended to limit the embodiments of the present invention. Other variations and modifications will be apparent to persons skilled in the art in light of the above description. And are neither required nor exhaustive of all embodiments. Any modification, equivalent replacement, and improvement made within the spirit and principle of the present invention should be included in the protection scope of the claims of the present invention.
Claims (8)
1. The parcel counting method based on the deep learning and multi-target tracking technology is characterized by comprising the following steps of:
s1, acquiring package data of logistics and preprocessing the package data; the logistics package data are divided into a training set and a verification set;
s2, training a preset deep learning-based package detection model through the training set, and testing and adjusting parameters of the trained package detection model through the verification set; obtaining a final package detection model;
s3, acquiring a real-time logistics package video, sampling key frames of the real-time logistics package video, and inputting the final package detection model for detection so as to obtain package position information of each key frame;
and S4, counting the packages by adopting a multi-target tracking algorithm according to the package position information of the key frames.
2. The method for counting parcels based on deep learning and multi-target tracking technology of claim 1, wherein the parcel data of the logistics of step S1 is obtained by arranging a camera device on a parcel conveyor belt and acquiring continuous parcel pictures according to a preset frequency.
3. The deep learning and multi-target tracking technology-based package counting method as claimed in claim 1, wherein the package data obtained in step S1 includes 10000 package pictures, 8000 package pictures are used as training set and 2000 package pictures are used as verification set.
4. The deep learning and multi-target tracking technology-based parcel counting method according to claim 3, wherein the preprocessing of step S1 is specifically: and marking the parcel boxes of the parcel pictures so as to obtain the position information of each parcel in each parcel picture, namely marking the coordinates of the upper left corner (x1, y1) and the coordinates of the lower right corner (x2, y2) of each parcel.
5. The deep learning and multi-target tracking technology-based package counting method according to claim 1, further comprising the following steps between the steps S1 and S2:
performing data enhancement processing including horizontal flipping, rotation and noise addition on the training set;
and inputting the training set into a preset generation countermeasure network for sample expansion.
6. The deep learning and multi-target tracking technology-based parcel counting method according to claim 1, wherein the logistics parcel video is acquired in real time by arranging a camera device on a parcel conveyor belt.
7. The deep learning and multi-target tracking technology-based parcel counting method according to claim 1, wherein the specific steps of said step S4 are:
obtaining a parcel corresponding to the parcel position information according to the parcel position information of the key frame, respectively assigning a unique id to each parcel detected before the current frame, and finding and matching each id and the parcel detected by the current frame by adopting a linear assignment algorithm:
for the successfully matched id, updating the current package position of the id to be the package position detected by the current frame;
for id which is not matched successfully, the parcel corresponding to the id enters an undetermined area and the position of the parcel is updated by adopting an LSTM algorithm or a Kalman filtering algorithm;
for the id which is not successfully matched in the multi-frame, if the tracked times exceed a preset threshold value, ending the id and increasing the number of packages by 1; and if the tracked times do not exceed the preset threshold value, determining that the detection is invalid, and not increasing the count.
8. The system for the package counting method based on the deep learning and multi-target tracking technology according to any one of claims 1 to 7, is characterized by comprising the following steps:
the data acquisition and pretreatment module is used for acquiring and pretreating the package data of the logistics; the logistics package data are divided into a training set and a verification set;
the package detection model training module is used for training a preset package detection model based on deep learning through the training set and testing and adjusting parameters of the trained package detection model through the verification set; obtaining a final package detection model;
the package detection module is used for acquiring a real-time logistics package video, sampling key frames of the real-time logistics package video, inputting the final package detection model for detection, and obtaining package position information of each key frame;
and the multi-target tracking parcel counting module is used for counting parcels by adopting a multi-target tracking algorithm according to the parcel position information of the key frame.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201911049870.7A CN110992305A (en) | 2019-10-31 | 2019-10-31 | Package counting method and system based on deep learning and multi-target tracking technology |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201911049870.7A CN110992305A (en) | 2019-10-31 | 2019-10-31 | Package counting method and system based on deep learning and multi-target tracking technology |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN110992305A true CN110992305A (en) | 2020-04-10 |
Family
ID=70082799
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201911049870.7A Pending CN110992305A (en) | 2019-10-31 | 2019-10-31 | Package counting method and system based on deep learning and multi-target tracking technology |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN110992305A (en) |
Cited By (8)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN111862036A (en) * | 2020-07-17 | 2020-10-30 | 中山大学 | Tool abnormal placement detection method and system based on double-finger scale quantity learning |
| CN113012136A (en) * | 2021-03-24 | 2021-06-22 | 中国民航大学 | Airport luggage counting method and counting system based on target detection |
| CN113610362A (en) * | 2021-07-20 | 2021-11-05 | 苏州超集信息科技有限公司 | Product tracing method and system based on deep learning assembly line |
| CN113807466A (en) * | 2021-10-09 | 2021-12-17 | 中山大学 | Logistics package autonomous detection method based on deep learning |
| CN114140735A (en) * | 2021-07-12 | 2022-03-04 | 苏州万店掌网络科技有限公司 | Deep learning-based goods path accumulation detection method and system and storage medium |
| CN115007483A (en) * | 2021-03-05 | 2022-09-06 | 法麦凯尼柯数据股份公司 | Method for dynamically analyzing packages |
| CN115035363A (en) * | 2022-05-30 | 2022-09-09 | 东南大学 | Online training method of deep neural network model |
| CN116109670A (en) * | 2022-12-31 | 2023-05-12 | 江苏中科贯微自动化科技有限公司 | Dynamic package counting method and system based on target detection |
Citations (11)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101141633A (en) * | 2007-08-28 | 2008-03-12 | 湖南大学 | Moving object detecting and tracing method in complex scene |
| CN108320297A (en) * | 2018-03-09 | 2018-07-24 | 湖北工业大学 | A kind of video object method for real time tracking and system |
| CN108711093A (en) * | 2018-05-18 | 2018-10-26 | 深圳市天趣控股有限公司 | A kind of intelligent contract method of lease in kind and system based on deep learning |
| CN108986064A (en) * | 2017-05-31 | 2018-12-11 | 杭州海康威视数字技术股份有限公司 | A kind of people flow rate statistical method, equipment and system |
| CN109448027A (en) * | 2018-10-19 | 2019-03-08 | 成都睿码科技有限责任公司 | A kind of adaptive, lasting motion estimate method based on algorithm fusion |
| CN109482507A (en) * | 2018-12-03 | 2019-03-19 | 安徽理工大学 | Count discriminate linear type express delivery sorting system |
| CN109522854A (en) * | 2018-11-22 | 2019-03-26 | 广州众聚智能科技有限公司 | A kind of pedestrian traffic statistical method based on deep learning and multiple target tracking |
| CN110021034A (en) * | 2019-03-20 | 2019-07-16 | 华南理工大学 | A kind of tracking recording broadcasting method and system based on head and shoulder detection |
| CN110084197A (en) * | 2019-04-28 | 2019-08-02 | 苏州清研微视电子科技有限公司 | Bus passenger flow volume statistical method and system based on deep learning |
| CN110163225A (en) * | 2018-02-11 | 2019-08-23 | 顺丰科技有限公司 | It is a kind of that package detection and recognition methods, apparatus and system are mixed based on cloud platform |
| CN209406858U (en) * | 2018-12-03 | 2019-09-20 | 安徽理工大学 | Count discriminate linear type express delivery sorting system |
-
2019
- 2019-10-31 CN CN201911049870.7A patent/CN110992305A/en active Pending
Patent Citations (11)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101141633A (en) * | 2007-08-28 | 2008-03-12 | 湖南大学 | Moving object detecting and tracing method in complex scene |
| CN108986064A (en) * | 2017-05-31 | 2018-12-11 | 杭州海康威视数字技术股份有限公司 | A kind of people flow rate statistical method, equipment and system |
| CN110163225A (en) * | 2018-02-11 | 2019-08-23 | 顺丰科技有限公司 | It is a kind of that package detection and recognition methods, apparatus and system are mixed based on cloud platform |
| CN108320297A (en) * | 2018-03-09 | 2018-07-24 | 湖北工业大学 | A kind of video object method for real time tracking and system |
| CN108711093A (en) * | 2018-05-18 | 2018-10-26 | 深圳市天趣控股有限公司 | A kind of intelligent contract method of lease in kind and system based on deep learning |
| CN109448027A (en) * | 2018-10-19 | 2019-03-08 | 成都睿码科技有限责任公司 | A kind of adaptive, lasting motion estimate method based on algorithm fusion |
| CN109522854A (en) * | 2018-11-22 | 2019-03-26 | 广州众聚智能科技有限公司 | A kind of pedestrian traffic statistical method based on deep learning and multiple target tracking |
| CN109482507A (en) * | 2018-12-03 | 2019-03-19 | 安徽理工大学 | Count discriminate linear type express delivery sorting system |
| CN209406858U (en) * | 2018-12-03 | 2019-09-20 | 安徽理工大学 | Count discriminate linear type express delivery sorting system |
| CN110021034A (en) * | 2019-03-20 | 2019-07-16 | 华南理工大学 | A kind of tracking recording broadcasting method and system based on head and shoulder detection |
| CN110084197A (en) * | 2019-04-28 | 2019-08-02 | 苏州清研微视电子科技有限公司 | Bus passenger flow volume statistical method and system based on deep learning |
Cited By (12)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN111862036A (en) * | 2020-07-17 | 2020-10-30 | 中山大学 | Tool abnormal placement detection method and system based on double-finger scale quantity learning |
| CN111862036B (en) * | 2020-07-17 | 2023-10-20 | 中山大学 | Tool abnormal placement detection method and system based on double-index measurement learning |
| CN115007483A (en) * | 2021-03-05 | 2022-09-06 | 法麦凯尼柯数据股份公司 | Method for dynamically analyzing packages |
| CN113012136A (en) * | 2021-03-24 | 2021-06-22 | 中国民航大学 | Airport luggage counting method and counting system based on target detection |
| CN114140735A (en) * | 2021-07-12 | 2022-03-04 | 苏州万店掌网络科技有限公司 | Deep learning-based goods path accumulation detection method and system and storage medium |
| CN113610362A (en) * | 2021-07-20 | 2021-11-05 | 苏州超集信息科技有限公司 | Product tracing method and system based on deep learning assembly line |
| CN113610362B (en) * | 2021-07-20 | 2023-08-08 | 苏州超集信息科技有限公司 | Deep learning assembly line-based product tracing method and system |
| CN113807466A (en) * | 2021-10-09 | 2021-12-17 | 中山大学 | Logistics package autonomous detection method based on deep learning |
| CN113807466B (en) * | 2021-10-09 | 2023-12-22 | 中山大学 | Logistics package autonomous detection method based on deep learning |
| CN115035363A (en) * | 2022-05-30 | 2022-09-09 | 东南大学 | Online training method of deep neural network model |
| CN115035363B (en) * | 2022-05-30 | 2024-04-16 | 东南大学 | Online training method of deep neural network model |
| CN116109670A (en) * | 2022-12-31 | 2023-05-12 | 江苏中科贯微自动化科技有限公司 | Dynamic package counting method and system based on target detection |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN110992305A (en) | Package counting method and system based on deep learning and multi-target tracking technology | |
| CN109840504B (en) | Article taking and placing behavior identification method and device, storage medium and equipment | |
| CN107527009B (en) | Remnant detection method based on YOLO target detection | |
| CN107609489A (en) | Calligraphy writing path evaluation device, method and electronic equipment | |
| CN110032916A (en) | A kind of method and apparatus detecting target object | |
| CN104613869B (en) | Method and system for detecting elliptical hole group based on maximum inscribed circle | |
| CN110287907A (en) | A kind of method for checking object and device | |
| CN109598298B (en) | Image object recognition method and system | |
| CN111597857B (en) | Logistics package detection method, device, equipment and readable storage medium | |
| CN106023249A (en) | Moving object detection method based on local binary similarity pattern | |
| CN106874913A (en) | A kind of vegetable detection method | |
| CN110458096A (en) | A kind of extensive commodity recognition method based on deep learning | |
| CN115619710A (en) | Method, device and equipment for counting quantity of articles based on step-by-step tracking detection | |
| CN114663390A (en) | Intelligent anti-pinch method, device, equipment and storage medium for automatic door | |
| CN114049568A (en) | Object shape change detection method, device, equipment and medium based on image comparison | |
| Wang et al. | Pointer meter recognition in UAV inspection of overhead transmission lines | |
| CN113781526A (en) | Domestic animal count identification system | |
| CN112153320B (en) | Method and device for measuring size of article, electronic equipment and storage medium | |
| CN108985325A (en) | A kind of classification method, device, equipment and system | |
| CN104349125B (en) | A kind of area monitoring method and equipment | |
| CN110782473A (en) | Conveyor belt static parcel detection method and detection system based on depth camera | |
| CN113378836A (en) | Image recognition method, apparatus, device, medium, and program product | |
| Zhihao et al. | Object detection algorithm based on dense connection | |
| CN114202041A (en) | Packaging material detection method and device | |
| CN114693735B (en) | Video fusion method and device based on target recognition |
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 |