Training a Custom model for Rear-Vehicle Detection with TensorFlow 2 Detection Model Zoo with bdd100k dataset.

Introduction about this repository:

• Read the descriptions and comments in Jupyter files.
• This repository is for Custom Rear-Vehicle Detection 'car, bus, truck' with bdd100k dataset
• The data preparation file depends on our use case, update the file to be compatible with your use case.
• We chose 'ssd_mobilenet_v2_fpnlite_320x320' for our case with trade-off its acceptable mean absolute precision 'mAP', and Speed in (ms) 'Problem dependant'.
• If you want to use another model, replace its name with our used model name, link, and model file name.

Warning: Make sure to Install TensorFlow with TensorFlow website

Some examples with test images :

Example with video :

Untitled.Project.mp4