This project consists of two componetns one is responsible for vehicle detection and tracking using Kalman filters and HOG+SVM and the other component is responsible for implementing lane detection Module 1(Detection consists of the following files) 1)desc.py responsible for computing HOG 2)detect.py main module which our main test file calls.This wraps the descriptor,sliding window ,kalman filter and threshold methods all in one file 3)scan.py performs sliding window detection at different scales 4)train.py performs training using SVm 5)track and match.py are modules responsible for performing kalman filter tracking and then selecting the most likely canidates for the bounding boxes detected by the detection module.For more details that underly the tracking concept please refer to https://arxiv.org/abs/1602.00763 The next module is the lane detection module. it consists of the following files. For an image obtained from the video,camera callibration and perspective mapping are performed. this is done by callibrate_cameras.py and perspective_transform.py Next we perform edge filtering and to get a better understanding of the lane markers we also augment the previous filtered result with an HLS filter.This is implemented in combined_thresh.py The modules responsible for performing line detection are implemented in lanefit.py Finally for visualization purposes we provide two files one for still image and one for video these are implemented in generate_examples.py and line_fit_video.py These algorithms were then tested on two videos challenge_video and project_video. The result can be viewed in challenge_video_result and project_video_result. Special Thanks to Udacity for introducing me to these wonderful concepts. Tracking parts some modules like finding correspondences between detected box and predicted box were implemented from scratch In Lane Finding module,line_fit and lanefit were implemented from scratch
sparshgarg23/object-detection-and-tracking-
Object detection and tracking in videos using HOG+SVM and Kalman Filter uses Filterpy,Numpy and Sci-kit learn
Python