iFinder: Structured Zero-Shot Vision-Based LLM Grounding for Dash-Cam Video Reasoning

• URL: http://arxiv.org/abs/2509.19552v2
• Date: Wed, 01 Oct 2025 06:54:44 GMT
• Title: iFinder: Structured Zero-Shot Vision-Based LLM Grounding for Dash-Cam Video Reasoning
• Authors: Manyi Yao, Bingbing Zhuang, Sparsh Garg, Amit Roy-Chowdhury, Christian Shelton, Manmohan Chandraker, Abhishek Aich,
• Abstract summary: iFinder is a semantic grounding framework that translates dash-cam videos into a hierarchical, interpretable data structure for large language models.<n>iFinder operates as a training-free pipeline that employs pretrained vision models to extract critical cues.<n>It significantly outperforms end-to-end V-VLMs on four zero-shot driving benchmarks.
• Score: 51.15353027471834
• License: http://arxiv.org/licenses/nonexclusive-distrib/1.0/
• Abstract: Grounding large language models (LLMs) in domain-specific tasks like post-hoc dash-cam driving video analysis is challenging due to their general-purpose training and lack of structured inductive biases. As vision is often the sole modality available for such analysis (i.e., no LiDAR, GPS, etc.), existing video-based vision-language models (V-VLMs) struggle with spatial reasoning, causal inference, and explainability of events in the input video. To this end, we introduce iFinder, a structured semantic grounding framework that decouples perception from reasoning by translating dash-cam videos into a hierarchical, interpretable data structure for LLMs. iFinder operates as a modular, training-free pipeline that employs pretrained vision models to extract critical cues -- object pose, lane positions, and object trajectories -- which are hierarchically organized into frame- and video-level structures. Combined with a three-block prompting strategy, it enables step-wise, grounded reasoning for the LLM to refine a peer V-VLM's outputs and provide accurate reasoning. Evaluations on four public dash-cam video benchmarks show that iFinder's proposed grounding with domain-specific cues, especially object orientation and global context, significantly outperforms end-to-end V-VLMs on four zero-shot driving benchmarks, with up to 39% gains in accident reasoning accuracy. By grounding LLMs with driving domain-specific representations, iFinder offers a zero-shot, interpretable, and reliable alternative to end-to-end V-VLMs for post-hoc driving video understanding.

Related papers

• LinkedOut: Linking World Knowledge Representation Out of Video LLM for Next-Generation Video Recommendation [32.57236582010967]
  Video Large Language Models (VLLMs) unlock world-knowledge-aware video understanding through pretraining on internet-scale data.<n>We present LinkedOut, a representation that extracts VLLM world knowledge directly from video to enable fast inference.<n>We introduce a cross-layer knowledge fusion MoE that selects the appropriate level of abstraction from the rich VLLM features, enabling personalized, interpretable, and low-latency recommendation.
  arXiv  Detail & Related papers  (2025-12-18T18:52:18Z)
• TRANSPORTER: Transferring Visual Semantics from VLM Manifolds [56.749972238005604]
  This paper introduces a logits-to-video (L2V) task alongside a model-independent approach, TRANSPORTER, to generate videos.<n> TRANSPORTER learns an optimal transport coupling to VLM's high-semantic embedding spaces.<n>In turn, logit scores define embedding directions for conditional video generation.
  arXiv  Detail & Related papers  (2025-11-23T09:12:48Z)
• LLM-RG: Referential Grounding in Outdoor Scenarios using Large Language Models [9.647551134303384]
  Referential grounding in outdoor driving scenes is challenging due to large scene variability, many visually similar objects, and dynamic elements.<n>We propose LLM-RG, a hybrid pipeline that combines off-the-shelf vision-language models for fine-grained attribute extraction with large language models for symbolic reasoning.
  arXiv  Detail & Related papers  (2025-09-29T21:32:54Z)
• Unleashing Hierarchical Reasoning: An LLM-Driven Framework for Training-Free Referring Video Object Segmentation [17.238084264485988]
  Referring Video Object (RVOS) aims to segment an object of interest throughout a video based on a language description.<n>bftextPARSE-VOS is a training-free framework powered by Large Language Models (LLMs)<n>bftextPARSE-VOS achieved state-of-the-art performance on three major benchmarks: Ref-YouTube-VOS, Ref-DAVIS17, and MeViS.
  arXiv  Detail & Related papers  (2025-09-06T15:46:23Z)
• VideoRefer Suite: Advancing Spatial-Temporal Object Understanding with Video LLM [81.15525024145697]
  Video Large Language Models (Video LLMs) have recently exhibited remarkable capabilities in general video understanding.<n>However, they mainly focus on holistic comprehension and struggle with capturing fine-grained spatial and temporal details.<n>We introduce the VideoRefer Suite to empower Video LLM for finer-level spatial-temporal video understanding.
  arXiv  Detail & Related papers  (2024-12-31T18:56:46Z)
• Harnessing Large Language Models for Training-free Video Anomaly Detection [34.76811491190446]
  Video anomaly detection (VAD) aims to temporally locate abnormal events in a video. Training-based methods are prone to be domain-specific, thus being costly for practical deployment. We propose LAnguage-based VAD (LAVAD), a method tackling VAD in a novel, training-free paradigm.
  arXiv  Detail & Related papers  (2024-04-01T09:34:55Z)
• Understanding Long Videos with Multimodal Language Models [44.78900245769057]
  Large Language Models (LLMs) have allowed recent approaches to achieve excellent performance on long-video understanding benchmarks.<n>We investigate how extensive world knowledge and strong reasoning skills of underlying LLMs influence this strong performance.<n>Our resulting Multimodal Video Understanding framework demonstrates state-of-the-art performance across multiple video understanding benchmarks.
  arXiv  Detail & Related papers  (2024-03-25T17:59:09Z)
• DoraemonGPT: Toward Understanding Dynamic Scenes with Large Language Models (Exemplified as A Video Agent) [73.10899129264375]
  This paper explores DoraemonGPT, a comprehensive and conceptually elegant system driven by LLMs to understand dynamic scenes.<n>Given a video with a question/task, DoraemonGPT begins by converting the input video into a symbolic memory that stores task-related attributes.<n>We extensively evaluate DoraemonGPT's effectiveness on three benchmarks and several in-the-wild scenarios.
  arXiv  Detail & Related papers  (2024-01-16T14:33:09Z)
• Divert More Attention to Vision-Language Object Tracking [87.31882921111048]
  We argue that the lack of large-scale vision-language annotated videos and ineffective vision-language interaction learning motivate us to design more effective vision-language representation for tracking. Particularly, in this paper, we first propose a general attribute annotation strategy to decorate videos in six popular tracking benchmarks, which contributes a large-scale vision-language tracking database with more than 23,000 videos. We then introduce a novel framework to improve tracking by learning a unified-adaptive VL representation, where the cores are the proposed asymmetric architecture search and modality mixer (ModaMixer)
  arXiv  Detail & Related papers  (2023-07-19T15:22:06Z)
• Dense Video Object Captioning from Disjoint Supervision [77.47084982558101]
  We propose a new task and model for dense video object captioning. This task unifies spatial and temporal localization in video. We show how our model improves upon a number of strong baselines for this new task.
  arXiv  Detail & Related papers  (2023-06-20T17:57:23Z)

This list is automatically generated from the titles and abstracts of the papers in this site.

This site does not guarantee the quality of this site (including all information) and is not responsible for any consequences.