TraVEL: Trajectory-Guided Video Embedding Learning for Driving-Video Retrieval
2026-08-13 • Computer Vision and Pattern Recognition
Computer Vision and Pattern RecognitionMachine Learning
AI summaryⓘ
The authors focus on improving how computer systems find specific driving moments, like turns or acceleration, in large video collections. They first fine-tune a general multimodal model with paired videos and descriptions, but this alone doesn't capture detailed motion well. To fix this, they develop TraVEL, which uses the car's movement paths during training to teach the model better motion understanding without needing extra data when searching. Their approach improves identifying motion-related video clips on a new benchmark compared to previous methods.
multimodal embeddingvideo retrievalInfoNCEego-trajectorypolicy optimizationfine-tuningdriving logsmotion understandingnuReasoningembedding-based search
Authors
Yi-Chung Chen, Philip Jacobson, Tom Lampo, Yiren Lu, Jin Yao, David I. Inouye, Jing Gao, Danhua Guo, Burhan Yaman
Abstract
Efficiently retrieving relevant clips from large-scale driving logs is essential for data curation, model development, and safety analysis. Structured and rule-based retrieval systems can explicitly target driving events, but typically require expert-defined rules, auxiliary data, and multi-stage perception pipelines. Multimodal embedding models offer a simpler and more efficient alternative by representing each video with a single searchable vector. However, general-purpose models often rely on shortcuts from static scene context and struggle to distinguish motion-centric events, such as turning left versus right or accelerating versus decelerating. In this work, we study how to adapt a general-purpose multimodal embedding model to driving-video retrieval. We first fine-tune Qwen3-VL-Embedding on paired clips and reasoning traces from nuReasoning using an InfoNCE objective. While this stage substantially improves overall retrieval, caption supervision alone remains insufficient for fine-grained motion understanding. We therefore introduce TraVEL (Trajectory-Guided Video Embedding Learning), a motion-aware fine-tuning framework that uses ego-trajectory similarity as a reward within Group Relative Policy Optimization. Trajectories serve only as privileged training supervision; retrieval still operates on single-vector video embeddings without ego poses, expert rules, or auxiliary perception outputs. We further construct a driving-video retrieval benchmark from nuReasoning. Experiments show that TraVEL improves motion-centric retrieval across model scales: relative to SFT, it raises longitudinal and lateral mAP by 9.8 and 4.7 points at 2B, with corresponding gains of 7.2 and 1.5 points at 8B. TraVEL thus combines physically grounded supervision with efficient embedding-based search.