Architectural Innovations for Efficient Denoising and Classification

Architectural Innovations for Efficient Denoising and Classification: A Manual vs. Neural Architecture Search Comparison

Markhorst, Thomas (TU Delft Electrical Engineering, Mathematics and Computer Science)

Kayhan, O.S. (mentor)
van Gemert, J.C. (mentor)
Demirović, E. (graduation committee)

Delft University of Technology

Computer Science | Artificial Intelligence

2023-08-22

In this paper, we combine image denoising and classification, aiming to enhance human perception of noisy images captured by edge devices, like security cameras. Since edge devices have little computational power, we also optimize for efficiency by proposing a novel architecture that integrates the two tasks. Additionally, we alter a Neural Architecture Search (NAS) method, which searches for classifiers, to search for the integrated model while optimizing for a target latency, classification accuracy, and denoising performance. Our NAS architectures outperform our manually designed alternatives in both denoising and classification, offering a significant improvement to human perception. Moreover, our approach empowers users to construct architectures tailored to domains like medical imaging, surveillance systems, and industrial inspections.

Denoising
Classification
Neural Architecture Search
Efficient
NAS
Noise
Image Quality
Latency

http://resolver.tudelft.nl/uuid:91cf78dd-7998-4322-9b99-f8ea7cf326c7

Student theses

master thesis

© 2023 Thomas Markhorst