See the past: Time-Reversed Scene Reconstruction from Thermal Traces Using Visual Language Models

2025-10-06

AI Summary

This work addresses the problem of reconstructing human activity scenes from residual thermal traces observed seconds to 120 seconds after occurrence. The proposed method introduces a *thermal trace time-reversal reconstruction framework*, leveraging RGB-thermal paired data and two collaborative vision-language models (VLMs) to generate semantic descriptions and structural priors, respectively—guiding a constrained diffusion process for semantically coherent thermal image reconstruction. Its key contribution is the first formulation of *reversible thermal trace inference*, modeling thermal decay as a time-reversible physical process and enforcing multimodal alignment to ensure reconstructed frames are physically plausible in semantics, geometry, and thermophysical properties. Evaluated across three controlled experimental settings, the system successfully recovers past-frame reconstructions consistent with heat diffusion laws and human behavioral priors. Results demonstrate both the temporal information richness embedded in thermal imagery and the feasibility of time-reversal thermal imaging.

Technology Category

• Computer Vision: Video Understanding & Activity Analysis
• Knowledge Representation and Reasoning: Geometric, Spatial, and Temporal Reasoning
• Planning, Routing, and Scheduling: Activity and Plan Recognition

Application Category

• Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for ranking
• Semantics and Knowledge: Data modeling to support human-machine intelligence, including LLMs agents, intelligent system behavior, explanations, and user-friendly interactions
• Graph Algorithms and Modeling for the Web: Representation, reconstruction, and subgraph or motif discovery in Web-related graphs

Abstract

Recovering the past from present observations is an intriguing challenge with potential applications in forensics and scene analysis. Thermal imaging, operating in the infrared range, provides access to otherwise invisible information. Since humans are typically warmer (37 C -98.6 F) than their surroundings, interactions such as sitting, touching, or leaning leave residual heat traces. These fading imprints serve as passive temporal codes, allowing for the inference of recent events that exceed the capabilities of RGB cameras. This work proposes a time-reversed reconstruction framework that uses paired RGB and thermal images to recover scene states from a few seconds earlier. The proposed approach couples Visual-Language Models (VLMs) with a constrained diffusion process, where one VLM generates scene descriptions and another guides image reconstruction, ensuring semantic and structural consistency. The method is evaluated in three controlled scenarios, demonstrating the feasibility of reconstructing plausible past frames up to 120 seconds earlier, providing a first step toward time-reversed imaging from thermal traces.

Problem

• Reconstruct past scene states from thermal traces
• Use paired RGB-thermal images for time-reversed recovery
• Infer recent human interactions through residual heat patterns

Innovation

• Time-reversed reconstruction from thermal traces
• Visual-Language Models guide constrained diffusion process
• Paired RGB-thermal images recover past scene states