An autoencoder with entropy loss, 4-bit quantization, and Huffman coding compresses in-sensor image features by up to 11985x, cutting estimated system energy by 2 to 4.5x with near-baseline accuracy on VWW, hand tracking, and eye tracking.
https://www.sony.com/en/SonyInfo/News/Press/ 202005/20-037E/, 2020
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
eess.IV 1years
2025 1verdicts
CONDITIONAL 1representative citing papers
citing papers explorer
-
OASIS: Optimized Lightweight Autoencoder System for Distributed In-Sensor computing
An autoencoder with entropy loss, 4-bit quantization, and Huffman coding compresses in-sensor image features by up to 11985x, cutting estimated system energy by 2 to 4.5x with near-baseline accuracy on VWW, hand tracking, and eye tracking.