Modojojo / traffic-surveillance-cars-proximity-detection

This repository contains code for car detection which also alerts whenever any of the cars are getting too close to each other, implemented using Tensorflow object detection

traffic-surveillance-cars-proximity-detection

This repository contains code for car detection which also alerts whenever any of the cars are getting too close to each other.

Repository Details

1. Contains the complete code for detecting cars, draw bounding boxes with class label and confidence and detect when cars are getting too close to each other.
2. Used pretrained faster-rcnn model from tensorflow model zoo and implemented wrapper function for ony vehicle detection.
3. The code correctly detects cars and applies a RED bounding boxes and 'TOO CLOSE' label on the cars whenever the get too close to each other.
4. 'steps_to_use.txt' file contains the steps required the run the 'application.py' file.

Remaining functionalities (To Add) :

1. Alert whenever the cars are getting too close
3. Re Train model on custom dataset for vehiles only

Example (GUI)

The fact that it is not able to detect some of the cars is because i used the pretrained model, i still have to train the models on the custom dataset so that it will be able to detect properly.