# First Place Solution to the ECCV 2024 ROAD++ Challenge @ ROAD++ Atomic Activity Recognition 2024

arXiv: https://arxiv.org/abs/2410.23092 · HTML: https://mlsanity.com/paper/2410.23092/

## Key facts

- No citing papers in the map yet (data through 14 Oct 2026).
- Builds on 5 of its 6 references that are in the map.
- First posted on arXiv 30 Oct 2024.

## Builds on (most cited first, 5 of 5)

- [Improved Regularization of Convolutional Neural Networks with Cutout](https://mlsanity.com/paper/1708.04552/) (2017)
- [X3D: Expanding Architectures for Efficient Video Recognition](https://mlsanity.com/paper/2004.04730/) (2020)
- [InternVideo: General Video Foundation Models via Generative and Discriminative Learning](https://mlsanity.com/paper/2212.03191/) (2022)
- [ROAD: The ROad event Awareness Dataset for Autonomous Driving](https://mlsanity.com/paper/2102.11585/) (2021)
- [Action-slot: Visual Action-centric Representations for Multi-label Atomic Activity Recognition in Traffic Scenes](https://mlsanity.com/paper/2311.17948/) (2023)

## Abstract (arXiv)

This report presents our team's technical solution for participating in Track 3 of the 2024 ECCV ROAD++ Challenge. The task of Track 3 is atomic activity recognition, which aims to identify 64 types of atomic activities in road scenes based on video content. Our approach primarily addresses the challenges of small objects, discriminating between single object and a group of objects, as well as model overfitting in this task. Firstly, we construct a multi-branch activity recognition framework that not only separates different object categories but also the tasks of single object and object group recognition, thereby enhancing recognition accuracy. Subsequently, we develop various model ensembling strategies, including integrations of multiple frame sampling sequences, different frame sampling sequence lengths, multiple training epochs, and different backbone networks. Furthermore, we propose an atomic activity recognition data augmentation method, which greatly expands the sample space by flipping video frames and road topology, effectively mitigating model overfitting. Our methods rank first in the test set of Track 3 for the ROAD++ Challenge 2024, and achieve 69% mAP.