Instance-Aware Knowledge Distillation for Semi-Supervised Learning of an On-Board Multi-Task Dense Prediction Model for Collision Avoidance System
2026-06-15 • Computer Vision and Pattern Recognition
Computer Vision and Pattern Recognition
AI summaryⓘ
The authors created a system that helps vehicles avoid collisions by using a small, fast neural network that can run on simple devices like those found in country clubs. They trained this smaller model using smart guesses from a bigger model combined with extra knowledge to reduce errors and the need for lots of labeled data. Their system identifies obstacles in front of the vehicle and sends information to help it move safely. Tests showed the smaller model is much more efficient and works better at segmenting objects compared to the bigger one, while still doing well at estimating depth.
collision avoidancedeep learningknowledge distillationsemi-supervised learninginstance segmentationmonocular depth estimationpseudo labelingedge computingautomated guided vehiclecontroller area network
Authors
Gyutae Hwang, Sang Jun Lee
Abstract
Collision avoidance systems have evolved toward camera-based deep learning approaches for driving scene understanding. However, deployment in edge environments such as country clubs is constrained by limited computational resources and unreliable communication infrastructure. Moreover, constructing large-scale datasets for the target domain involves substantial annotation cost. To address these limitations, we propose an instance-aware knowledge distillation framework for semi-supervised learning. Specifically, we generate pseudo labels that mitigate teacher bias by leveraging domain priors from the teacher and instance-centric knowledge from foundation models. The trained lightweight student is deployed in the proposed collision avoidance system and performs multiple dense prediction tasks in real-time. The system detects frontal obstacles and encodes their spatial information into controller area network messages for automated guided vehicle operation. To achieve this, we construct a large-scale country club dataset and perform field validation of the proposed system. Experimental results demonstrate that the student outperforms the large teacher in instance segmentation while mitigating performance degradation in monocular depth estimation. Compared with the teacher, the student reduces FLOPs by 22.68$\times$ and parameters by 14.33$\times$, achieving 6.46 FPS on a low-cost edge device.