Evaluation and Statistical Analysis Code for "Multi-Task Learning for Joint Fisheye Compression and Perception for Autonomous Driving"

<p dir="ltr">This repository contains the Python scripts used for the evaluation and statistical analysis in the paper "Multi-Task Learning for Joint Fisheye Compression and Perception for Autonomous Driving".</p><p dir="ltr">The code includes:</p&g...

Full description

Saved in:
Bibliographic Details
Main Author: Basem Ahmed (18127861) (author)
Published: 2025
Subjects:
Tags: Add Tag
No Tags, Be the first to tag this record!
Description
Summary:<p dir="ltr">This repository contains the Python scripts used for the evaluation and statistical analysis in the paper "Multi-Task Learning for Joint Fisheye Compression and Perception for Autonomous Driving".</p><p dir="ltr">The code includes:</p><ul><li><a target="_blank"><code>evaluate_mtl.py</code></a>: The main script for evaluating the performance of the proposed deep learning models (JointGAD) and traditional codecs (HEVC, JPEG2000) on the Woodscape and Fisheye8k datasets. It calculates metrics for image compression (BPP, PSNR, MS-SSIM) and perception (mAP, mIoU), and saves per-class results for statistical analysis.</li><li><code>perform_ttest.py</code>: A script to perform a paired t-test on the per-class metrics generated by the evaluation script. This is used to verify that the observed performance differences between models are statistically supported. The script can also generate a LaTeX table of the results.</li></ul><p dir="ltr">These scripts are implemented in Python using the PyTorch framework and are provided to ensure the reproducibility of the experimental results presented in the manuscript.</p>