# SemanticSegmentation

**Repository Path**: zhaoyewei/SemanticSegmentation

## Basic Information

- **Project Name**: SemanticSegmentation
- **Description**: No description available
- **Primary Language**: Unknown
- **License**: Not specified
- **Default Branch**: master
- **Homepage**: None
- **GVP Project**: No

## Statistics

- **Stars**: 0
- **Forks**: 0
- **Created**: 2021-01-19
- **Last Updated**: 2021-01-19

## Categories & Tags

**Categories**: Uncategorized

**Tags**: None

## README

# Semantic Segmentation

Semantic Segmentation on CityScapes dateset using PyTorch.

CityScapes dataset structure should be:
data_dir/cityscapes/leftImg8bit
data_dir/cityscapes/gtFine

Steps to run - 
1. run utils/preprocessdata.py
2. run train.ipynb
3. run evaluation/eval_on_val.ipynb
4. run evaluation/eval_on_test.ipynb

model/resnet.py:
Definition of the custom Resnet model (output stride = 8 or 16)

model/aspp.py:
Definition of the Atrous Spatial Pyramid Pooling (ASPP) module.

utils/preprocess_data.py:
Converts all Cityscapes label images from having Id to having trainId pixel values, and saves these to data_dir/cityscapes/meta/label_imgs. Also computes class weights.