First Place Solution to the ECCV 2024 ROAD++ Challenge @ ROAD++ Spatiotemporal Agent Detection 2024

TL;DR

This paper presents a comprehensive solution for spatiotemporal agent detection in videos, addressing challenges like object size, low-light conditions, and class imbalance, achieving first place in the ECCV 2024 ROAD++ Challenge.

Contribution

The authors introduce novel detection heads, a dual-stream model with low-light enhancement, and a multi-branch framework with optimized training strategies for improved agent detection.

Findings

Achieved 30.82% average video-mAP on the test set.

Ranked first in the ECCV 2024 ROAD++ Challenge.

Enhanced detection of extreme-size objects and low-light scenes.

Abstract

This report presents our team's solutions for the Track 1 of the 2024 ECCV ROAD++ Challenge. The task of Track 1 is spatiotemporal agent detection, which aims to construct an "agent tube" for road agents in consecutive video frames. Our solutions focus on the challenges in this task, including extreme-size objects, low-light scenarios, class imbalance, and fine-grained classification. Firstly, the extreme-size object detection heads are introduced to improve the detection performance of large and small objects. Secondly, we design a dual-stream detection model with a low-light enhancement stream to improve the performance of spatiotemporal agent detection in low-light scenes, and the feature fusion module to integrate features from different branches. Subsequently, we develop a multi-branch detection framework to mitigate the issues of class imbalance and fine-grained classification, and we design a pre-training and fine-tuning approach to optimize the above multi-branch framework. Besides, we employ some common data augmentation techniques, and improve the loss function and upsampling operation. We rank first in the test set of Track 1 for the ROAD++ Challenge 2024, and achieve 30.82% average video-mAP.