This repository contains all the necessary tools and instructions to replicate the studies and experiments presented in our paper: "Multi-Objective Optimization with Homotopy-Based Strategies for Enhanced Multimodal Automatic Target Recognition Models" authored by Sophia Abraham, Steve Cruz, Suya You, Jonathan D. Hauenstein, and Walter Scheirer.
Automatic Target Recognition (ATR) often confronts intricate visual scenes, necessitating models capable of discerning subtle nuances. This project utilizes a homotopy-based multi-objective optimization technique with the CLIP model to enhance multimodal model precision and generalizability for ATR applications. The integration of vision and language model CLIP and sophisticated optimization strategies exemplifies the potential to significantly boost performance on complex ATR tasks.
