As an undergraduate researcher, I'm building foundational expertise in machine learning and its applications. My current focus lies in computer vision, particularly exploring novel approaches to image quality assessment without reference images.
My broader interests span automation, computational methods, and the integration of machine learning with physics and hardware systems.
Proposes a dynamic routing pipeline for No-Reference Image Quality Assessment that avoids full-ensemble execution. A lightweight gradient-boosted classifier selects the optimal IQA model from a diverse pool of traditional (BRISQUE, NIQE) and deep-learning (MANIQA, HyperIQA) algorithms based on low-cost perceptual features. This "predict-then-execute" approach achieves performance competitive with the best individual deep model (HyperIQA) at a fraction of the computational cost of full-ensemble execution.
DOI: 10.1109/ETFI68128.2026.11484766 · ORCID: 0009-0009-2696-2519
@INPROCEEDINGS{11484766,
author={Maddel, Arya and Patil, Ratnajeet and Bhavsar, Tiya and Gajbahar, Anisha and Jaiswal, Swati and Joshi, Prachi},
booktitle={2026 International Conference on Emerging Technologies and Future Innovations (ETFI)},
title={Adaptive Hybrid Architecture for NRIQA},
year={2026},
pages={1-7},
doi={10.1109/ETFI68128.2026.11484766},
month={Feb},
}
Awarded first place in the "AI For Humanity" competition organized by DES Pune University in collaboration with IEEE Pune Section and its Computer Society, I&M Society, and Student Branch chapters.
Secured the first prize in the Inter Department Project Competition (NIRMITEE) at Sou. Venutai Chavan Polytechnic, Sinhagad Institutes.