This project aims to combine stereo vision with an object ranging algorithm (YOLO) and add in object recognition.
Clone the GitHub repository. Warning: the dataset is large (~2GB). Navigate to stereo_vision_for_object_ranging.py and edit the master_path_to_dataset variable to be a string of the path to the dataset in your system. Change the skip_forward_file_pattern to a time in order to skip to that specific time.
The default approach is dense stereo with all its pre and postprocessing and without WLS filtering - To switch to WLS Filtering with dense stereo, set the "WLS_on" variable to True (line 19) - To switch to sparse stereo, set the "sparse_ORB" variable to True (line 20) - To switch back to dense stereo with no WLS filtering, set both variables to False (lines 19, 20) - IMPORTANT: DO NOT SET BOTH TO TRUE AT THE SAME TIME!
Run the program by running stereo_vision_for_object_ranging.py either in the command prompt or from an IDE of your choice, and sit back and watch the program recognise objects and distances!