Goal: Definitely one of the more difficult and open-ended issues. Please look at the rr_ml to see what has currently been done. Currently, it using a segmentation NN, and the code works but it is a bit slow, if you can increase the speed then that would be great. We will probably run this on a Jetson Nano and the main priorities will be speed and accuracy. Here are some labeled data if you plan to train the NN.
Files Edited (at least): ml/*
Success Criteria:
CNN that detects lane lines at least at 10-50 Hz so we can use it for real-time applications.
Goal: Definitely one of the more difficult and open-ended issues. Please look at the
rr_mlto see what has currently been done. Currently, it using a segmentation NN, and the code works but it is a bit slow, if you can increase the speed then that would be great. We will probably run this on a Jetson Nano and the main priorities will be speed and accuracy. Here are some labeled data if you plan to train the NN.Files Edited (at least): ml/*
Success Criteria: