CN208479837U - A kind of real-time target detection device based on raspberry pie - Google Patents

A kind of real-time target detection device based on raspberry pie Download PDF

Info

Publication number
CN208479837U
CN208479837U CN201821181582.8U CN201821181582U CN208479837U CN 208479837 U CN208479837 U CN 208479837U CN 201821181582 U CN201821181582 U CN 201821181582U CN 208479837 U CN208479837 U CN 208479837U
Authority
CN
China
Prior art keywords
raspberry pie
development board
video data
real
camera
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.)
Expired - Fee Related
Application number
CN201821181582.8U
Other languages
Chinese (zh)
Inventor
李彤岩
肖翔
徐嘉临
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Chengdu University of Information Technology
Original Assignee
Chengdu University of Information Technology
Priority date (The priority date 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 date listed.)
Filing date
Publication date
Application filed by Chengdu University of Information Technology filed Critical Chengdu University of Information Technology
Priority to CN201821181582.8U priority Critical patent/CN208479837U/en
Application granted granted Critical
Publication of CN208479837U publication Critical patent/CN208479837U/en
Expired - Fee Related legal-status Critical Current
Anticipated expiration legal-status Critical

Links

Landscapes

  • Image Analysis (AREA)

Abstract

The utility model belongs to target detection technique field, discloses a kind of real-time target detection device based on raspberry pie, and the real-time target detection device based on raspberry pie is provided with raspberry pie development board;Raspberry pie development board is connected with power supply by 3.3v or 5v interface, raspberry pie development board is connected with camera by CSI connector, the HDMI interface of raspberry pie development board is connected with display by HDMI connection VGA patchcord, and the USB interface of raspberry pie development board is connected with keyboard and mouse.The utility model acquires video data by camera, these data can be uploaded to cloud by network by raspberry pie development board, and user can be checked by intelligent terminal.The raspberry pie development board that the utility model is used has the advantage that small in size for other processing boards, structure is simple, suitable for the development and utilization of video object detection, can effectively real-time perfoming target detection, reach better video monitoring effect.

Description

A kind of real-time target detection device based on raspberry pie
Technical field
The utility model belongs to target detection technique field more particularly to a kind of real-time target detection dress based on raspberry pie It sets.
Background technique
Currently, the prior art commonly used in the trade is this bar:
Raspberry pie development board is a kind of minicomputer that research and development are supported by Britain's raspberry pie development board foundation, and size is only Have bank card size, it is cheap, enter in terms of study or scientific research gate threshold tool it is not high, have good compatibility.Tree The certain kind of berries sends development board not only to come with simple computer operating system, also has very strong Video coding and decoding capability;Therefore it sets The certain kind of berries send development board exploitation can be used in multiple fields, such as: text-to-speech service, FM station transmitter, Hadoop cluster, Video object detection and tracking etc..
With the quickening of Development of Urbanization, many cells can all arrange monitoring device, but the cost that monitoring device is high And the key that monitoring device can not be popularized.
Video object detection technique is mainly by time in computer vision methods subduction video and redundancy spatially Information has certain sensitivity to target position variation.Current this video object detection technique mainly applies to intelligent prison In control or target identification and tracking.But the hsrdware requirements for meeting above-mentioned technical proposal in industry are too high, especially X86 chip It is all expensive with GPU, cause cost too high;Limited transmission distance system;Software opening is poor;Phenomena such as power consumption is higher.
In conclusion problem of the existing technology is:
Existing video object detection device is too high to hardware requirement, causes cost too high;Limited transmission distance system;It is soft Part opening is poor;Phenomena such as power consumption is higher.
Utility model content
In view of the problems of the existing technology, the utility model provides a kind of real-time target detection dress based on raspberry pie It sets.
The utility model is that this bar is realized, a kind of real-time target detection device based on raspberry pie is provided with
Raspberry pie development board;
Raspberry pie development board is connected with power supply by 3.3v or 5v interface, and raspberry pie development board is connected by CSI connector There is camera, the HDMI interface of raspberry pie development board is connected with display, raspberry pie development board by HDMI connection VGA patchcord USB interface be connected with keyboard and mouse;
Raspberry pie development board includes video data acquiring device, video data processor, moving object detection device and video Data storage;
Camera mode in raspberry pie development board is configured to video input mode first by video data acquiring device, then Video data is acquired using V4L2 technology;
Video data processor is after the information for receiving camera acquisition, at 64 ARMV8 of raspberry pie development board Coded treatment quickly is carried out to video information after managing module, by video data encoding collected at h.264 format;
Moving object detection device builds the convolution mind an of lightweight using MXNet inside raspberry pie exploitation electrode systems Computer vision system is created through network, and on raspberry pie development board;
In video data memory, mainly by being connected on AWS Cloud, is managed using cloud and carry out lightweight convolution The management of neural network and the storage of data.
Further, raspberry pie development board is Raspberry Pi3 Model B, is four core Broadcom of a new generation 64 ARMv8 processors of BCM2837, processor speed can reach 1.2GHz, and ARMv8 includes the memory of 1GB RAM;4 USB interface, 40 pin GPIO plugs of expansion, onboard BCM43143WiFi chip support bluetooth and WIFI;HDMI output end Mouthful;Storage device: microSD;SD card card reader.
Further, camera using Camera V2.1 model camera.
Further, camera is connected by CSI connector.
Further, keyboard is connect with mouse by USB interface.
Further, cable interface or onboard WiFi chip are connected to the network.
Further, electric source line interface connects Android charging cable, voltage 3.3v or 5v.
Further, HDMI output port connection display institute.
Further, the convolutional neural networks model of the raspberry pie Development plank system that microSD storage has been downloaded and training.
The utility model acquires video data by camera, and detecting the rate that it identifies target is 2.067s/ frame, average Precision reaches 71.24mAP, and storage of the whole system from video data acquiring to training result can reach real-time well These data are finally uploaded to cloud by network using raspberry pie development board, user can pass through intelligent end by the purpose of property End is checked.Whole system is all significantly less than monitoring system on the market in cost and operation cost, and this is practical new The raspberry pie development board that type is used has the advantage that small in size for other processing boards, and structure is simple, is applicable in In the development and utilization of video object detection, can effectively real-time perfoming target detection, reach better video monitoring effect.
Detailed description of the invention
Fig. 1 is the real-time target structure of the detecting device schematic diagram provided by the embodiment of the utility model based on raspberry pie;
Fig. 2 is the real-time target detection device working principle diagram provided by the embodiment of the utility model based on raspberry pie;
Fig. 3 is the hardware structure diagram of raspberry pie development board provided by the embodiment of the utility model;
Fig. 4 is V4L2 video capture technology schematic diagram provided by the embodiment of the utility model;
In figure: 1, raspberry pie development board;2, power supply;3, camera;4, display;5, keyboard;6, mouse.
Specific embodiment
For the invention, features and effects that can further appreciate that the utility model, the following examples are hereby given, and cooperates Detailed description are as follows for attached drawing.
As depicted in figs. 1 and 2, the real-time target detection device packet provided by the embodiment of the utility model based on raspberry pie It includes: raspberry pie development board 1, power supply 2, camera 3, display 4, keyboard 5, mouse 6.
Raspberry pie development board 1 is connected with power supply 2 by 3.3v or 5v interface, and raspberry pie development board 1 is connected by CSI connector It is connected to camera 3, the HDMI interface of raspberry pie development board 1 is connected with display 4, raspberry pie by HDMI connection VGA patchcord The USB interface of development board 1 is connected with keyboard 5 and mouse 6;
Raspberry pie development board 1 includes video data acquiring device, video data processor, moving object detection device and video Data storage;
Video data acquiring device is that the camera mode in raspberry pie development board 1 is configured to video input mode;
Video data processor is after the information for receiving the acquisition of camera 3, at 64 ARMV8 of raspberry pie development board Quickly video information is handled after reason module;
Moving object detection device builds the convolution of a lightweight using MXNet in 1 internal system of raspberry pie development board Neural network, and computer vision system is created on raspberry pie development board 1;
Video data memory is managed using cloud mainly by being connected on AWS Cloud and carries out lightweight convolution mind The storage of management and data through network.
System Working Principle: video data acquiring device master, which plays, connects the completion of camera 3 by CSI connector, first will tree The certain kind of berries sends the camera mode of the camera 3 of development board 1 to be configured to video input mode.After the completion of configuration, regarded by Linux V4L2 Frequency acquisition technique is acquired video data;Video data processor is after receiving collected video data, by video Data are encoded, and coded format is h.264 format;Moving object detection device builds one by MXNet deep learning library A lightweight volume convolutional neural networks, the video data encoded is put into convolutional neural networks and is trained, is obtained Training result;Video data memory, will be trained as a result, by being connected on AWS Cloud, utilize cloud management to carry out The management of lightweight convolutional neural networks and the storage of data.
Further, raspberry pie development board 1 is 3 Model B of Raspberry Pi, is four core Broadcom of a new generation 64 ARMv8 processors of BCM2837, processor speed can reach 1.2GHz, and ARMv8 includes the memory of 1GB RAM;4 USB interface, 40 pin GPIO plugs of expansion, onboard BCM43143 WiFi chip support bluetooth and WIFI;HDMI output end Mouthful;Storage device: microSD;SD card card reader.
Further, camera 3 using Camera V2.1 model camera.
When the utility model is used, the camera 3 transferred is fixed on somewhere, whenever someone passes by its camera 3 When institute's coverage area, video data at this time can be passed through by the internal system that camera 3 is passed to raspberry pie development board 1 ARMV8 processor handles video data.The video data of incoming 1 internal system of raspberry pie development board can be supported a variety of Storage format be connected to above display 4 including H264, mp4 etc. by HDMI, controlled at the end PC.It is calculated using SSD Method is realized in 1 inner utilization MXNet deep learning library of raspberry pie development board, builds the convolutional Neural net an of lightweight Network, the video data being passed to camera 3 are trained, and carry out detection judgement to the target of be passed to video data, In the region covered, if someone occurs, detection and tracking can be carried out to it, and data at this time are stored, and passes through It allows and sends out standby and create secure connection in AWS Cloud;AWS loT can be used after connection to create on raspberry pie development board 1 Service, it can near-real-time carries out target detection and result is pushed in AWS Cloud.It can also send out and set at this time It detects and target detection video data at this time is sent to user by way of mail or short message while target.
The above is only the preferred embodiment to the utility model, is not made in any form to the utility model Limitation, it is all according to the technical essence of the utility model any simple modification made to the above embodiment, equivalent variations with Modification, is all within the scope of the technical scheme of the utility model.

Claims (3)

1. a kind of real-time target detection device based on raspberry pie, which is characterized in that the real-time target inspection based on raspberry pie Device is surveyed to be provided with
Raspberry pie development board;
Raspberry pie development board is connected with power supply by 3.3v or 5v interface, and raspberry pie development board is connected with by CSI connector and is taken the photograph As head, the HDMI interface of raspberry pie development board is connected with display by VGA patchcord, and the USB interface of raspberry pie development board connects It is connected to keyboard and mouse;
Raspberry pie development board includes video data acquiring device, video data processor, moving object detection device and video data Memory;
Video data acquiring device is that the camera mode in raspberry pie development board is configured to video input mode;
Video data processor handles mould after the information for receiving camera acquisition, through 64 ARMV8 of raspberry pie development board Quickly video information is handled after block;
Moving object detection device builds the convolutional neural networks of a lightweight inside raspberry pie Development plank system, and is setting The certain kind of berries, which is sent, creates computer vision system on development board;
In video data memory, by being connected on AWS Cloud, is managed using cloud and carry out lightweight convolutional neural networks Management and data storage.
2. the real-time target detection device based on raspberry pie as described in claim 1, which is characterized in that raspberry pie development board is adopted With BCM283764 ARMv8 processors of four core Broadcom, ARMv8 includes the memory of 1GB RAM;4 USB interfaces expand The 40 pin GPIO plugs filled, onboard BCM43143WiFi chip support bluetooth and WIFI;HDMI output port;Storage device: microSD;SD card card reader.
3. the real-time target detection device based on raspberry pie as described in claim 1, which is characterized in that camera using The camera of Camera V2.1 model.
CN201821181582.8U 2018-07-25 2018-07-25 A kind of real-time target detection device based on raspberry pie Expired - Fee Related CN208479837U (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN201821181582.8U CN208479837U (en) 2018-07-25 2018-07-25 A kind of real-time target detection device based on raspberry pie

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN201821181582.8U CN208479837U (en) 2018-07-25 2018-07-25 A kind of real-time target detection device based on raspberry pie

Publications (1)

Publication Number Publication Date
CN208479837U true CN208479837U (en) 2019-02-05

Family

ID=65208997

Family Applications (1)

Application Number Title Priority Date Filing Date
CN201821181582.8U Expired - Fee Related CN208479837U (en) 2018-07-25 2018-07-25 A kind of real-time target detection device based on raspberry pie

Country Status (1)

Country Link
CN (1) CN208479837U (en)

Cited By (8)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN109981378A (en) * 2019-04-16 2019-07-05 陈麒任 A kind of network detection and analysis tool and its application method based on raspberry pie
CN110110639A (en) * 2019-04-29 2019-08-09 济南浪潮高新科技投资发展有限公司 A kind of Indoor Video method and Indoor Video vehicle based on artificial intelligence tracking
CN110472541A (en) * 2019-08-05 2019-11-19 福州大学 Garment identification system and method based on raspberry pie
CN110569827A (en) * 2019-09-28 2019-12-13 华南理工大学 Face recognition reminding system based on convolutional neural network
CN110609225A (en) * 2019-11-08 2019-12-24 伟创力电子技术(苏州)有限公司 PCBA general test platform based on raspberry group
CN111251299A (en) * 2020-02-21 2020-06-09 广东工业大学 Hadoop-based cleaning type cloud robot system
CN112309068A (en) * 2020-10-29 2021-02-02 电子科技大学中山学院 Forest fire early warning method based on deep learning
CN113743242A (en) * 2021-08-13 2021-12-03 成都理工大学 Landslide mass detection and identification method based on raspberry pie

Cited By (9)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN109981378A (en) * 2019-04-16 2019-07-05 陈麒任 A kind of network detection and analysis tool and its application method based on raspberry pie
CN110110639A (en) * 2019-04-29 2019-08-09 济南浪潮高新科技投资发展有限公司 A kind of Indoor Video method and Indoor Video vehicle based on artificial intelligence tracking
CN110472541A (en) * 2019-08-05 2019-11-19 福州大学 Garment identification system and method based on raspberry pie
CN110569827A (en) * 2019-09-28 2019-12-13 华南理工大学 Face recognition reminding system based on convolutional neural network
CN110569827B (en) * 2019-09-28 2024-01-05 华南理工大学 Face recognition reminding system based on convolutional neural network
CN110609225A (en) * 2019-11-08 2019-12-24 伟创力电子技术(苏州)有限公司 PCBA general test platform based on raspberry group
CN111251299A (en) * 2020-02-21 2020-06-09 广东工业大学 Hadoop-based cleaning type cloud robot system
CN112309068A (en) * 2020-10-29 2021-02-02 电子科技大学中山学院 Forest fire early warning method based on deep learning
CN113743242A (en) * 2021-08-13 2021-12-03 成都理工大学 Landslide mass detection and identification method based on raspberry pie

Similar Documents

Publication Publication Date Title
CN208479837U (en) A kind of real-time target detection device based on raspberry pie
CN103297754B (en) A kind of monitor video self-adaptation encoding region of interest system
CN102546338B (en) Multimedia intelligent sensor network system and method based on controller area network (CAN) bus
CN203909869U (en) Cloud door control barrier gate management system
CN103604273B (en) The method of intelligent refrigerator food control
CN107766812B (en) MiZ 702N-based real-time face detection and recognition system
CN102891796A (en) Mining intelligent transmission gateway
CN110751117A (en) Unmanned aerial vehicle-based sheep flock and cattle flock quantity monitoring method and device
CN114374709A (en) 5G video and Internet of things distribution network monitoring system and method based on edge cloud cooperation
CN101720024A (en) Intelligent recognition monitoring system and monitoring method thereof
CN105447647A (en) Shield intelligent maintenance management system
CN102625082A (en) Video monitoring system
CN108765674A (en) A kind of access control system and method
CN103678522A (en) Method for acquiring and converting data of metering system of intelligent transformer substation
CN104853150A (en) Multi-camera objective cooperatively tracking technology
CN207720304U (en) A kind of intelligent remote video monitoring system merged based on mobile intelligent terminal and technology of Internet of things
CN207232656U (en) A kind of intelligent control well site monitoring device
CN104182818B (en) A kind of method and system of substation information intelligence automatic data collection
CN104463114A (en) Method for catching images and quickly recognizing targets and embedded device
CN104463143A (en) Identification system based on image processing
CN112218044A (en) Analysis system and analysis method with security inspection artificial intelligence recognition function
CN208798111U (en) A kind of recognition of face cloud talk-back host
CN104301409B (en) A kind of information processing system and method
CN207717137U (en) A kind of computer network security monitoring device
CN205827474U (en) A kind of gate with intelligent recognition function

Legal Events

Date Code Title Description
GR01 Patent grant
GR01 Patent grant
CF01 Termination of patent right due to non-payment of annual fee

Granted publication date: 20190205

Termination date: 20190725

CF01 Termination of patent right due to non-payment of annual fee