Road objects detection

Training detection models (RetinaNet and SSD) to detect road objects, then applying a model to real world traffic video from Moscow.

Detection dataset

For model training the Traffic Road Object Detection Polish 12k was used.

Usage

• The first option is to open and run the notebook /notebooks/road_objects_detection.ipynb with comments and visualizations in Kaggle or Google Colab.

• The second option is cloning the repo, installing the needed requirements, and working locally:

git clone https://github.com/RadyaSRN/road-objects-detection.git
cd road-objects-detection
conda create -n roadobj python=3.10
conda activate roadobj
pip install -r requirements.txt