Distortion Convolution Module for Semantic Segmentation: Code Implementation and Evaluation

This repository provides the code for the Distortion Convolution Module for Semantic Segmentation, as presented in our IEEE Transactions on Image Processing (TIM) paper. The code is implemented in PyTorch and requires the following packages: PyTorch, tqdm, and TensorboardX.

To train and evaluate the model, use the following command:

python train.py --batch-size 8 --lr 0.007 

This will initiate the training process and provide both training and testing results.

For using custom datasets or panoramic images, ensure to update the camera's intrinsic parameters within the sperical.py file located in the modeling/s2cnn directory.