Semantic-Segmentation-with-ENet

Tensorflow 2.0 Implementation of the E-Net Semantic Segmentation Architecture

Based on the paper "ENet: A Deep Neural Network Architecture for Real-Time Semantic Segmentation"

Setup

• Clone the repository
• Install the package locally using the following command

pip install -e Semantic-Segmentation-with-Enet/

Files in the Repository

• tests/tests_enet.py : Python script to test the dimensions of all the blocks and model using random data

• evaluation/dataset_prep.py : Prepares the training data and decodes & encodes the segmented images in the dataset

• enet_seg/enet_blocks.py : Definitions of the Bottleneck layers in the architecture

• enet_seg/enet_model.py : Final model including all the layers outlined in the paper

• enet_seg/utilities.py : Contains the definition of the Max Unpool layer