Streaming Video Object Detection and Tracking: Raspberry Pi, Python, openCV
Budget: $750 – $1,500 USD
Streaming Video Object Detection and Tracking: Raspberry Pi, Python, openCV
$750.00 – 1,500.00 USD
Hello Raspberry Pi Developers!
PLEASE READ this description very carefully. Take your time, I will not decide on a freelancer within hours. This is an AI camera detection project . If you are not expert in this field, please skip.
PLEASE READ the attached PDF document to this proposal. It has much more detail than this description.
ONLY APPLY IF you are fluent in Python on a Raspberry Pi. Must have exposure to openVC and using RTSP video streams from IP cameras.
MUST HAVE a Raspberry Pi 4 single board computer.
Scope of Project
To develop a vision-based autonomous system to detect migrating birds from the video stream of a thermal IP Camera. The video stream will be taken from the RTSP (Real-Time Streaming Protocol) of the IP camera. We have the camera and will provide you with a link to the RTSP video stream (winning bid).
For testing purposes you will need to work from saved video files that are located on our YouTube channel over the RTSP video stream since birds are not migrating currently. Links below to video clips of birds migrating in the 2022 fall season taken from the thermal IP camera.
The birds are small and at high altitude, and will appear and just dots on the screen (only a few pixels wide). We are looking for a developer to develop a Raspberry Pi 4 web server application to detect the individual birds from the RTSP video stream of the IP camera, count them, track the entry x,y position in the frame, and track the exit x,y position as it exits the frame, and store this data along with the time of detection in a database.
Example video footage of migrating birds taken from the IP thermal camera:
https://www.youtube.com/watch?v=vcNefjWtV4s
Development Framework and Strategy
The framework for the system should be web based. We should be able to access the application from a browser. Consider running the Pi code on something like Caddy Web Server. This is a Phase 1 development application for proof on concept. If we can get this performing well we will award a Phase 2 development for porting this code to a Cloud Web Server. Please keep this in mind developing this application.
Tasks to be Completed
1. Pull the RTSP video stream from the Thermal IP camera into openCV to be analyzed. (The link to the RTSP camera will be provide to the developer awarded this bid) Note: This should be a parameter we can change easily for different camera RTSP links.
2. Develop a live video detection algorithm using openCV to detect each bird coming into the frame and track the bird until it exits the frame. Since the birds are just dots you can use the openCV blob detection which should make this much simpler.
3. For each detection record the detection in a database which includes the x,y pixel position on the video frame that the bird enters the frame along with the time & date, and the x,y pixel position the and time & date the bird exits the frame.
4. Display the live video feed with boxes around the bird detected as they move across the frame.
5. Histogram display of detection data using something like Plotly.js, https://plotly.com/. Include a histogram of total detection's per day and another graph showing total detection per hour of each individual day.
6. Include a method to download the contents of the database by a date range in CSV format.
Deliverables
1. The source code will be put into a private GitHub repository. Login information to the private repository will be give upon the award of this project.
2. A Raspberry Pi image will be created that can easily be installed on to a SD card by using the Raspberry Pi Imager or balenEtcher.
$750.00 – 1,500.00 USD
Hello Raspberry Pi Developers!
PLEASE READ this description very carefully. Take your time, I will not decide on a freelancer within hours. This is an AI camera detection project . If you are not expert in this field, please skip.
PLEASE READ the attached PDF document to this proposal. It has much more detail than this description.
ONLY APPLY IF you are fluent in Python on a Raspberry Pi. Must have exposure to openVC and using RTSP video streams from IP cameras.
MUST HAVE a Raspberry Pi 4 single board computer.
Scope of Project
To develop a vision-based autonomous system to detect migrating birds from the video stream of a thermal IP Camera. The video stream will be taken from the RTSP (Real-Time Streaming Protocol) of the IP camera. We have the camera and will provide you with a link to the RTSP video stream (winning bid).
For testing purposes you will need to work from saved video files that are located on our YouTube channel over the RTSP video stream since birds are not migrating currently. Links below to video clips of birds migrating in the 2022 fall season taken from the thermal IP camera.
The birds are small and at high altitude, and will appear and just dots on the screen (only a few pixels wide). We are looking for a developer to develop a Raspberry Pi 4 web server application to detect the individual birds from the RTSP video stream of the IP camera, count them, track the entry x,y position in the frame, and track the exit x,y position as it exits the frame, and store this data along with the time of detection in a database.
Example video footage of migrating birds taken from the IP thermal camera:
https://www.youtube.com/watch?v=vcNefjWtV4s
Development Framework and Strategy
The framework for the system should be web based. We should be able to access the application from a browser. Consider running the Pi code on something like Caddy Web Server. This is a Phase 1 development application for proof on concept. If we can get this performing well we will award a Phase 2 development for porting this code to a Cloud Web Server. Please keep this in mind developing this application.
Tasks to be Completed
1. Pull the RTSP video stream from the Thermal IP camera into openCV to be analyzed. (The link to the RTSP camera will be provide to the developer awarded this bid) Note: This should be a parameter we can change easily for different camera RTSP links.
2. Develop a live video detection algorithm using openCV to detect each bird coming into the frame and track the bird until it exits the frame. Since the birds are just dots you can use the openCV blob detection which should make this much simpler.
3. For each detection record the detection in a database which includes the x,y pixel position on the video frame that the bird enters the frame along with the time & date, and the x,y pixel position the and time & date the bird exits the frame.
4. Display the live video feed with boxes around the bird detected as they move across the frame.
5. Histogram display of detection data using something like Plotly.js, https://plotly.com/. Include a histogram of total detection's per day and another graph showing total detection per hour of each individual day.
6. Include a method to download the contents of the database by a date range in CSV format.
Deliverables
1. The source code will be put into a private GitHub repository. Login information to the private repository will be give upon the award of this project.
2. A Raspberry Pi image will be created that can easily be installed on to a SD card by using the Raspberry Pi Imager or balenEtcher.