Object detection updates sent efficiently to edge traffic systems
MECAIL: Communication-Aware Incremental Learning for Object Detection with 14.6 KB Spatiotemporal Experts
Computer Vision and Pattern RecognitionMachine Learning
Summary
Many smart traffic devices need to keep getting better by learning new things, but they don’t have enough power to learn everything themselves. The authors created tiny add-on modules that help these devices adapt to specific places like parking lots or gas stations. These add-ons are very small—only 14.6 KB—so they can be quickly sent over usual wireless networks without problems. Their approach performs nearly as well as bigger, more complex methods but uses much less data, making it practical for real-world use.
What this means in practice
- •For traffic system engineers: Install small expert modules to update object detection models in edge devices for specific road environments without heavy data transfer.
- •For mobile network operators: Support efficient transmission of incremental learning modules over existing mobile networks to enhance edge device capabilities.
Authors
Matthias Neuwirth-Trapp, Maarten Bieshaar, Danda Paudel, Konrad Schindler, Luc Van Gool, Christos Sakaridis
Abstract
Intelligent transportation systems require Incremental Learning (IL) to continually improve their overall performance in dynamic environments. However, most edge devices lack the computational resources to support on-device IL, requiring updates to be transmitted from centralized servers. We propose using this setup to obtain dense, specialized module coverage that adapts a fixed base model to specific spatiotemporal contexts, such as parking lots, gas stations, ferries, or construction sites. However, in order to reliably transmit these modules to the edge device, using TCP, UDP, and BTP over V2X, Wi-Fi, and 2G-5G hardware, we establish a strict limit of 14.6 KB per module to fit within the first TCP window and to minimize UDP/BTP fragmentation. We further introduce Mixture-of-Experts for Communication-Aware Incremental Learning (MECAIL), the first method that meets this strict requirement, in which each new domain or environment is served by a small expert network that adapts the base model. We validate MECAIL on D-RICO and ODinW-13, where it largely matches the performance of parameter-heavy approaches while enabling practical, bandwidth-efficient large-scale deployment. This allows comprehensive coverage by experts for highly specific, focused, and temporary situations.