用三阶段流水线实现零样本交通事故理解，性能远超基线。
Multi-Stage VLM Pipeline for Zero-Shot Traffic Accident Understanding
- 构建三阶段流水线：全视频联合预测、时间精修、单帧撞击点定位
- 在公开/私有榜单分别获0.55469/0.57080，较最强基线提升约0.21
- 适合关注零样本事故分析与多模型融合的自动驾驶研究者
我们提交了在CVPR 2026 AUTOPILOT Workshop ACCIDENT挑战赛中获得第一名的解决方案，该任务要求从CCTV视频中进行零样本事故时间、撞击中心点和碰撞类型的预测。基于冻结的Qwen3-VL-32B-Instruct检查点，我们构建了一个三阶段流水线（全视频联合预测、时间精修、单帧撞击中心点定位），并在一个235B参数的Mixture-of-Experts兄弟模型上再次运行该流水线，以9:1比例融合两者输出，并将每个预测点投影到最近的车辆检测框上。最终系统在公开榜单达0.55469，私有榜单达0.57080，较最强基线（Molmo-7B）的0.358提升约0.21，成功夺冠。我们对各组件进行了消融实验，报告了影响设计的关键负结果，并在https://github.com/fuumin621/cvpr2026-accident-1st-place-solution开源代码。
原文摘要 · Abstract (English)
We present the 1st-place solution to the ACCIDENT challenge at the CVPR 2026 AUTOPILOT Workshop, which asks for zero-shot prediction of accident timing, impact centroid, and collision type from CCTV footage. On a frozen Qwen3-VL-32B-Instruct checkpoint we build a three-stage pipeline (full-video joint prediction, time refinement, and single-frame grounding of the impact centroid), run the same pipeline a second time on a 235B Mixture-of-Experts sibling, blend the two outputs 9:1, and finally snap each predicted point onto the nearest vehicle detection. The final system reaches Public LB 0.55469 / Private LB 0.57080, roughly +0.21 over the strongest host baseline (Molmo-7B, 0.358) and wins the challenge. We ablate each component, report the negative results that shaped the final design, and release the code at https://github.com/fuumin621/cvpr2026-accident-1st-place-solution.
Thank you to arXiv for use of its open access interoperability. PaperDance 不是 arXiv 官方产品；中文卡片由大模型生成，请以原文为准。