Chimera: A Block-Based Neural Architecture Search Framework for Event-Based Object Detection
This work addresses the challenge of efficiently designing neural networks for event-based cameras, which is incremental as it builds on existing NAS and event processing techniques.
The paper tackled the problem of adapting RGB-domain processing methods to event-based object detection by proposing Chimera, a block-based neural architecture search framework, which achieved performance comparable to state-of-the-art models on the PEDRo dataset with an average parameter reduction of 1.6 times.
Event-based cameras are sensors that simulate the human eye, offering advantages such as high-speed robustness and low power consumption. Established Deep Learning techniques have shown effectiveness in processing event data. Chimera is a Block-Based Neural Architecture Search (NAS) framework specifically designed for Event-Based Object Detection, aiming to create a systematic approach for adapting RGB-domain processing methods to the event domain. The Chimera design space is constructed from various macroblocks, including Attention blocks, Convolutions, State Space Models, and MLP-mixer-based architectures, which provide a valuable trade-off between local and global processing capabilities, as well as varying levels of complexity. The results on the PErson Detection in Robotics (PEDRo) dataset demonstrated performance levels comparable to leading state-of-the-art models, alongside an average parameter reduction of 1.6 times.