- This Machine Learning Project has been made for demonstrating Machine Learning Operations (MLOps) for Semantic Segmentation on BDD100K dataset using Weights and Biases dashboard. The dataset contains thousands of images of self driving cars.
- The aim of this project is to perceive the object in front of the car, be it traffic light, road or a person.
- The model is then optimized through hyperparameter tuning and evaluated using individual IOU (Intersection over Union) scores. The results of the model training, optimization and evaluation through individual reports, along with the details of job and sweep runs are given at: Weights and Biases Workspace.
Along with .py files, the Python notebooks as alternatives to local GPU have been provided so that they can be run in Google Colab or Kaggle.
Additionally, the .pth file for this Machine Learning model can be accessed at: Hugging Face directory since it was too large to be uploaded at this GitHub repository.