Train a YOLO v4 Neural Network to detect motorcycle odometers from video footage
Budget: $30 – $250 USD
I would like someone to train a YOLO v4 neural network to do the following:
- Detect Motorcycle odometers in Video footage
- Detect borders of the dashboard and highlight them within a box
- Be able to extract the border boxes from each frame in the video footage
- Should be able to detect any kind of dashboard shape or type whether digital or analogue
- Should be able to accurately blur the boxes found by the detector
I will provide a video sample that is used for testing the neural network once it's trained and ready. The network should be able to highlight every frame as attached sample image.
The work needs to provide as an outcome a fully trained neural network that can be run on any footage to detect the dashboard and highlight it with an output file for the bounding box of the dashboard for each frame.
Source code should be provided as well as explanation on how the work was done and how to train for additional objects.
- Detect Motorcycle odometers in Video footage
- Detect borders of the dashboard and highlight them within a box
- Be able to extract the border boxes from each frame in the video footage
- Should be able to detect any kind of dashboard shape or type whether digital or analogue
- Should be able to accurately blur the boxes found by the detector
I will provide a video sample that is used for testing the neural network once it's trained and ready. The network should be able to highlight every frame as attached sample image.
The work needs to provide as an outcome a fully trained neural network that can be run on any footage to detect the dashboard and highlight it with an output file for the bounding box of the dashboard for each frame.
Source code should be provided as well as explanation on how the work was done and how to train for additional objects.