Reproducibility of "FDA: Fourier Domain Adaptation forSemantic Segmentation

TL;DR

This paper provides a detailed reproducibility report for the FDA method in semantic segmentation, including code, ablation studies, and implementation details to verify the original results.

Contribution

It offers a comprehensive reproducibility analysis of FDA, including code, ablations, and instructions, enhancing transparency and replicability in semantic segmentation research.

Findings

Reproduced original FDA results successfully

Provided detailed ablation studies and implementation instructions

Confirmed the effectiveness of Fourier domain adaptation

Abstract

The following paper is a reproducibility report for "FDA: Fourier Domain Adaptation for Semantic Segmentation" published in the CVPR 2020 as part of the ML Reproducibility Challenge 2020. The original code was made available by the author. The well-commented version of the code containing all ablation studies performed derived from the original code along with WANDB integration is available at <github.com/thefatbandit/FDA> with proper instructions to execute experiments in README.