Summary:

This is an application of YOLO based on Redmon et al., 2016 and Redmon and Farhadi, 2016 for Car Detection. Here we are implementing non-max supression, intersectio over union, YOLO evaluation and handing bounding box annotation. The dataset used here is from drive.ai.

Here we built the model from scratch but used pretrained weights for speeding up the training. We also implemented Intersection Over Union and Non-max supression for improving accuracy as well as provided some code for running the model on a list of images and converting it to video.