Vehicle Detection using Mask R-CNN and Computer Vision based Lane Detection
Implemented the Mask R-CNN using Keras and TensorFlow. The model detects vehicles in the image frame using segmentation masks with the pretrained weights trained on COCO dataset; the lane detection is done using sobel filter. This project is a part of Udacity Self-Driving Car Engineer program
Result image frames:
Mask R-CNN output on a road scene:
Land Identification output:
Final combined output:
Download the pretrained weights here and place it in the current working directory. Run main.py specifying the input image path. It can be applied with video files using moviepy and calling process_video() function.
Final Video GIF: