Abhishek Nair’s recent work in NeuRonICS Lab, IISc (PI: Chetan Singh Thakur) jointly with Prof. Shantanu Chakrabartty, published in high impact factor journal IEEE Internet of things.
This work presents a novel in-filter computing framework that can be used for designing ultra-light acoustic classifiers for use in smart internet-of-things (IoTs). Unlike a conventional acoustic pattern recognizer, where the feature extraction and classification are designed independently, the proposed architecture integrates the convolution and nonlinear filtering operations directly into the kernels of a Support Vector Machine (SVM). The result of this integration is a template-based SVM whose memory and computational footprint (training and inference) is light enough to be implemented on an FPGA-based IoT platform. While the proposed in-filter computing framework is general enough, in this paper, we demonstrate this concept using a Cascade of Asymmetric Resonator with Inner Hair Cells (CAR-IHC) based acoustic feature extraction algorithm. We show that the system can achieve robust classification performance on benchmark sound recognition tasks using only ~1.5k Look-Up Tables (LUTs) and ~2.8k Flip-Flops (FFs), a significant improvement over other approaches.
The development of the same algorithm has been demonstrated in real-time by Hemanth and Vishnu on TI’s development board CC1352R for bird detection with low memory and power footprint of 20KB SRAM and 14mW respectively.
Hemanth Sabbella demonstrated a system using this framework at tinyML Asia 2021:
Bird Hotspots: A tinyML acoustic classification system for ecological insights
The proposed system is demonstrated in real-time to detect and log bird species and their occurrences running on ARM Cortex M4 processor consuming 1.6mA of mean current. It is estimated to last for at least 2 months for detecting multiple bird species on 3 AA batteries and could be optimized to achieve backup durations up to 1-year. The logged data is used to make chronological hotspots of bird occurrences on google maps which could help understand valuable information towards species conservation.
Bird Density Identification using AudioMoth
The green light blinking on AudioMoth denotes the identification of cuckoo bird sound.