/bdd100k-instance-segmentation

Primary LanguagePythonOtherNOASSERTION

BDD100k Instance Segmentation with Mask R-CNN

This is a final project for a masters in data science at the University of Denver. Instance Segmentation with Berkeley DeepDrive dataset using Mask R-CNN on Python 3, Keras, and TensorFlow. The model generates bounding boxes and segmentation masks for each instance of an object in the image. It's based on Feature Pyramid Network (FPN) and a ResNet50 backbone.

Instance Segmentation Sample

Getting Started

  • eda.ipynb shows some statistical analysis on the bdd100k ins seg dataset.

  • data_inspect.ipynb shows inspection of the BDD100k dataset and Mask R-CNN model.

  • model_training.ipynb It shows the Berkeley DeepDrive dataset trained for instance segmentation with Mask R-CNN

  • data_inspect.ipynb shows evaluation of the Mask R-CNN model on the BDD100k dataset.

Requirements

Python 3.4, TensorFlow 1.3, Keras 2.0.8 and other common packages listed in requirements.txt.