arXiv subjects

Lalithkumar Seenivasan

Publications and source records attributed to Lalithkumar Seenivasan.

At least 19 recordsLinked to original sources

arXiv2026

Dense Structural Priors for Sparse Functional Landmark Localization in Surgical Videos

Vision foundation models such as SAM 3 can provide transferable object-level structure across diverse surgical video conditions, but segmentation outputs do not explicitly encode the action-conditioned semantics that define functional surgical landmarks. Estimating instrument extent and geometry differs from localizing the tip or anchor relevant to clipping, grasping, or dissecting. We investigate vision foundation model-enabled sparse action-aware landmark localization, using zero-shot, point-prompted structural masks to provide dense instrument-level context without manual pixel-level mask annotations. We propose a lightweight refinement framework that uses SAM 3 as a structural prior. A coarse multi-frame network predicts tip and anchor prompts, generating non-oracle masks that are fused with visual and heatmap features to refine functional landmark predictions. We compare direct mask-augmented supervision, prediction-derived mask-prior refinement, and auxiliary mask supervision to examine how vision foundation model-derived structure should enter a precision-oriented localization system. Experiments on 7,867 clips from 60 surgical videos spanning YouTube, Cholec80, HeiChole, SurgVU, and CRCD evaluate the approach under heterogeneous conditions. Without manual pixel-level mask annotations for training, the proposed model achieves overall F1 scores of 72.4% for tip and 58.0% for anchor localization. Ablations show that coarse-to-fine refinement provides a substantial performance gain, while prediction-derived structural priors provide additional improvement when incorporated as intermediate guidance rather than direct localization targets.

cs.CV↗

arXiv2026

C3VD-DEFCOL: A Deformable Colonoscopy Dataset with Time-Resolved 3D Ground Truth and Realistic Appearance

3D reconstruction could improve colonoscopy by estimating mucosal coverage and alerting clinicians to missed regions during screening. However, algorithm development is limited as no current datasets provide both a realistic in vivo appearance and dense, time-resolved 3D ground truth, especially under non-rigid deformation. We present C3VD-DEFCOL, a framework and dataset for evaluating deformable colonoscopy reconstruction with paired geometry and realistic texture. Starting from C3VD/C3VDv2 colon meshes and camera trajectories, we generate controlled deformations of the colon surface, including peristaltic waves and centerline motion, and render per-frame depth, surface normals, optical flow, camera poses, and time-stamped 3D meshes. We then use the rendered geometry, primarily depth, to condition an LTX-2.3-based sim-to-real translation model that produces RGB clips with in vivo-like mucosal color, texture, vasculature, and specular appearance while preserving the underlying 3D scene structure. The resulting dataset contains 110 videos from 11 unique colon mesh geometries, with varying camera trajectories, appearances, and parameterized deformation regimes, including three peristaltic severity levels that serve as controlled evaluation axes. We evaluate the generated videos using appearance realism, geometric consistency, and temporal consistency metrics, and use the paired ground truth to benchmark the downstream task of pose estimation in deformable 3D reconstruction. Our experiments show how pose estimation error increases with increasing deformation severity, providing a controlled stress test that is not possible with existing in vivo datasets. Overall, C3VD-DEFCOL is designed as a reproducible, quantitative evaluation platform for testing deformable 3D reconstruction algorithms, with the goal of reducing the domain gap between synthetic datasets and in vivo colonoscopy.

cs.CV↗

arXiv2026

Investigating Robot Control Policy Learning for Autonomous X-ray-guided Spine Procedures

Imitation learning-based robot control policies are enjoying renewed interest in video-based robotics. However, it remains unclear whether this approach applies to X-ray-guided procedures, such as spine instrumentation, with sparse inputs. We examine the feasibility, opportunities and challenges for imitation policy learning in bi-plane-guided cannula insertion. We develop an in silico sandbox for scalable, automated simulation of X-ray-guided spine procedures with a high degree of realism. We curate a dataset of correct trajectories and corresponding bi-planar X-ray sequences that emulate the stepwise alignment of providers. We then train imitation learning policies for planning and open-loop control that iteratively align a cannula in a vertebroplasty setting solely based on visual information. This precisely controlled setup offers insights into limitations and capabilities of this method. Our policy succeeded on the first attempt in 68.5% of cases, maintaining safe intra-pedicular trajectories across diverse vertebral levels. The policy transferred to complex anatomy, including fractures, as well as varied anatomies and initializations. Rollouts on real X-ray indicate that partial sim-to-real transfer with plausible trajectories is possible. While these preliminary results are promising, we also identify limitations, especially in entry point precision. The current results present a clear benchmark for future efforts, while with more robust priors and domain knowledge, such models may provide a foundation for future efforts toward lightweight and CT-free robotic intra-operative spinal navigation.

cs.CV↗

arXiv2026

SegSTRONG-C: Segmenting Surgical Tools Robustly On Non-adversarial Generated Corruptions -- An EndoVis'24 Challenge

Surgical data science has seen rapid advancement with the excellent performance of end-to-end deep neural networks (DNNs). Despite their successes, DNNs have been proven susceptible to minor "corruptions," introducing a major concern for the translation of cutting-edge technology, especially in high-stakes scenarios. We introduce the SegSTRONG-C challenge dedicated to better understanding model deterioration under unforeseen but plausible non-adversarial "corruption" and the capabilities of contemporary methods that seek to improve it. Built on a dataset generated through counterfactual robotic replay, SegSTRONG-C provides paired clean and "corrupted" samples, enabling reproducible evaluation of model robustness. Participants are challenged to train tool segmentation algorithms on "uncorrupted" data and evaluate them on "corrupted" test domains for the binary robot tool segmentation task. Through comprehensive baseline experiments and participating submissions from widespread community engagement, SegSTRONG-C reveals key themes for model failure and identifies promising directions for improving robustness. The performance of challenge winners, achieving an average 0.9394 DSC and 0.9301 NSD across the unreleased test sets with "corruption" types: bleeding, smoke, and low brightness. This highlights how prior knowledge, customized training strategies, and architectural choice can be leveraged to improve robustness. In conclusion, the SegSTRONG-C challenge has identified practical approaches for enhancing model robustness. However, most approaches rely on conventional techniques that have known limitations. Looking ahead, we advocate for expanding intellectual diversity and creativity in non-adversarial robustness beyond data augmentation, calling for new paradigms that enhance universal robustness to unforeseen "corruptions" to facilitate richer applications in surgical data science.

cs.CV↗

arXiv2026

Towards Robust Surgical Automation via Digital Twin Representations from Foundation Models

Large language model-based (LLM) agents are emerging as a powerful enabler of robust embodied intelligence due to their capability of planning complex action sequences. Sound planning ability is necessary for robust automation in many task domains, but especially in surgical automation. These agents rely on a highly detailed natural language representation of the scene. Thus, to leverage the emergent capabilities of LLM agents for surgical task planning, developing similarly powerful and robust perception algorithms is necessary to derive a detailed scene representation of the environment from visual input. Previous research has focused primarily on enabling LLM-based task planning while adopting simple yet severely limited perception solutions to meet the needs for bench-top experiments, but lacks the critical flexibility to scale to less constrained settings. In this work, we propose an alternate perception approach -- a digital twin (DT)-based machine perception approach that capitalizes on the convincing performance and out-of-the-box generalization of recent vision foundation models. Integrating our DT representation and LLM agent for planning with the dVRK platform, we develop an embodied intelligence system and evaluate its robustness in performing peg transfer and gauze retrieval tasks. Our approach shows strong task performance and generalizability to varied environmental settings. Despite a convincing performance, this work is merely a first step towards the integration of DT representations. Future studies are necessary for the realization of a comprehensive DT framework to improve the interpretability and generalizability of embodied intelligence in surgery.

cs.RO↗

arXiv2026

TwinOR: Photorealistic Digital Twins of Dynamic Operating Rooms for Embodied AI Research

Developing embodied AI for intelligent surgical systems requires safe, controllable environments for continual learning and evaluation. However, safety regulations and operational constraints in operating rooms (ORs) limit agents from freely perceiving and interacting in realistic settings. Digital twins provide high-fidelity, risk-free environments for exploration and training. How we may create dynamic digital representations of ORs that capture relevant spatial, visual, and behavioral complexity remains an open challenge. We introduce TwinOR, a real-to-sim infrastructure for constructing photorealistic and dynamic digital twins of ORs. The system reconstructs static geometry and continuously models human and equipment motion. The static and dynamic components are fused into an immersive 3D environment that supports controllable simulation and facilitates future embodied exploration. The proposed framework reconstructs complete OR geometry with centimeter-level accuracy while preserving dynamic interaction across surgical workflows. In our experiments, TwinOR synthesizes stereo and monocular RGB streams as well as depth observations for geometry understanding and visual localization tasks. Models such as FoundationStereo and ORB-SLAM3 evaluated on TwinOR-synthesized data achieve performance within their reported accuracy ranges on real-world indoor datasets, demonstrating that TwinOR provides sensor-level realism sufficient for emulating real-world perception and localization challenge. By establishing a perception-grounded real-to-sim pipeline, TwinOR enables the automatic construction of dynamic, photorealistic digital twins of ORs. As a safe and scalable environment for experimentation, TwinOR opens new opportunities for translating embodied intelligence from simulation to real-world clinical environments.

cs.CV↗

arXiv2026

AffordTissue: Dense Affordance Prediction for Tool-Action Specific Tissue Interaction

Surgical action automation has progressed rapidly toward achieving surgeon-like dexterous control, driven primarily by advances in learning from demonstration and vision-language-action models. While these have demonstrated success in table-top experiments, translating them to clinical deployment remains challenging: current methods offer limited predictability on where instruments will interact on tissue surfaces and lack explicit conditioning inputs to enforce tool-action-specific safe interaction regions. Addressing this gap, we introduce AffordTissue, a multimodal framework for predicting tool-action specific tissue affordance regions as dense heatmaps during cholecystectomy. Our approach combines a temporal vision encoder capturing tool motion and tissue dynamics across multiple viewpoints, language conditioning enabling generalization across diverse instrument-action pairs, and a DiT-style decoder for dense affordance prediction. We establish the first tissue affordance benchmark by curating and annotating 15,638 video clips across 103 cholecystectomy procedures, covering six unique tool-action pairs involving four instruments (hook, grasper, scissors, clipper) and their associated tasks: dissection, grasping, clipping, and cutting. Experiments demonstrate substantial improvement over vision-language model baselines (20.6 px ASSD vs. 60.2 px for Molmo-VLM), showing that our task-specific architecture outperforms large-scale foundation models for dense surgical affordance prediction. By predicting tool-action specific tissue affordance regions, AffordTissue provides explicit spatial reasoning for safe surgical automation, potentially unlocking explicit policy guidance toward appropriate tissue regions and early safe stop when instruments deviate outside predicted safe zones.

cs.CV↗

arXiv2026

Investigating a Policy-Based Formulation for Endoscopic Camera Pose Recovery

In endoscopic surgery, surgeons continuously locate the endoscopic view relative to the anatomy by interpreting the evolving visual appearance of the intraoperative scene in the context of their prior knowledge. Vision-based navigation systems seek to replicate this capability by recovering camera pose directly from endoscopic video, but most approaches do not embody the same principles of reasoning about new frames that makes surgeons successful. Instead, they remain grounded in feature matching and geometric optimization over keyframes, an approach that has been shown to degrade under the challenging conditions of endoscopic imaging like low texture and rapid illumination changes. Here, we pursue an alternative approach and investigate a policy-based formulation of endoscopic camera pose recovery that seeks to imitate experts in estimating trajectories conditioned on the previous camera state. Our approach directly predicts short-horizon relative motions without maintaining an explicit geometric representation at inference time. It thus addresses, by design, some of the notorious challenges of geometry-based approaches, such as brittle correspondence matching, instability in texture-sparse regions, and limited pose coverage due to reconstruction failure. We evaluate the proposed formulation on cadaveric sinus endoscopy. Under oracle state conditioning, we compare short-horizon motion prediction quality to geometric baselines achieving lowest mean translation error and competitive rotational accuracy. We analyze robustness by grouping prediction windows according to texture richness and illumination change indicating reduced sensitivity to low-texture conditions. These findings suggest that a learned motion policy offers a viable alternative formulation for endoscopic camera pose recovery.

eess.IV↗

arXiv2026

SAW: Toward a Surgical Action World Model via Controllable and Scalable Video Generation

A surgical world model capable of generating realistic surgical action videos with precise control over tool-tissue interactions can address fundamental challenges in surgical AI and simulation -- from data scarcity and rare event synthesis to bridging the sim-to-real gap for surgical automation. However, current video generation methods, the very core of such surgical world models, require expensive annotations or complex structured intermediates as conditioning signals at inference, limiting their scalability. Other approaches exhibit limited temporal consistency across complex laparoscopic scenes and do not possess sufficient realism. We propose Surgical Action World (SAW) -- a step toward surgical action world modeling through video diffusion conditioned on four lightweight signals: language prompts encoding tool-action context, a reference surgical scene, tissue affordance mask, and 2D tool-tip trajectories. We design a conditional video diffusion approach that reformulates video-to-video diffusion into trajectory-conditioned surgical action synthesis. The backbone diffusion model is fine-tuned on a custom-curated dataset of 12,044 laparoscopic clips with lightweight spatiotemporal conditioning signals, leveraging a depth consistency loss to enforce geometric plausibility without requiring depth at inference. SAW achieves state-of-the-art temporal consistency (CD-FVD: 199.19 vs. 546.82) and strong visual quality on held-out test data. Furthermore, we demonstrate its downstream utility for (a) surgical AI, where augmenting rare actions with SAW-generated videos improves action recognition (clipping F1-score: 20.93% to 43.14%; cutting: 0.00% to 8.33%) on real test data, and (b) surgical simulation, where rendering tool-tissue interaction videos from simulator-derived trajectory points toward a visually faithful simulation engine.

cs.CV↗

arXiv2026

Towards Controllable Video Synthesis of Routine and Rare OR Events

Purpose: Curating large-scale datasets of operating room (OR) workflow, encompassing rare, safety-critical, or atypical events, remains operationally and ethically challenging. This data bottleneck complicates the development of ambient intelligence for detecting, understanding, and mitigating rare or safety-critical events in the OR. Methods: This work presents an OR video diffusion framework that enables controlled synthesis of rare and safety-critical events. The framework integrates a geometric abstraction module, a conditioning module, and a fine-tuned diffusion model to first transform OR scenes into abstract geometric representations, then condition the synthesis process, and finally generate realistic OR event videos. Using this framework, we also curate a synthetic dataset to train and validate AI models for detecting near-misses of sterile-field violations. Results: In synthesizing routine OR events, our method outperforms off-the-shelf video diffusion baselines, achieving lower FVD/LPIPS and higher SSIM/PSNR in both in- and out-of-domain datasets. Through qualitative results, we illustrate its ability for controlled video synthesis of counterfactual events. An AI model trained and validated on the generated synthetic data achieved a RECALL of 70.13% in detecting near safety-critical events. Finally, we conduct an ablation study to quantify performance gains from key design choices. Conclusion: Our solution enables controlled synthesis of routine and rare OR events from abstract geometric representations. Beyond demonstrating its capability to generate rare and safety-critical scenarios, we show its potential to support the development of ambient intelligence models.

cs.CV↗

arXiv2025

DualVision ArthroNav: Investigating Opportunities to Enhance Localization and Reconstruction in Image-based Arthroscopy Navigation via External Cameras

Arthroscopic procedures can greatly benefit from navigation systems that enhance spatial awareness, depth perception, and field of view. However, existing optical tracking solutions impose strict workspace constraints and disrupt surgical workflow. Vision-based alternatives, though less invasive, often rely solely on the monocular arthroscope camera, making them prone to drift, scale ambiguity, and sensitivity to rapid motion or occlusion. We propose DualVision ArthroNav, a multi-camera arthroscopy navigation system that integrates an external camera rigidly mounted on the arthroscope. The external camera provides stable visual odometry and absolute localization, while the monocular arthroscope video enables dense scene reconstruction. By combining these complementary views, our system resolves the scale ambiguity and long-term drift inherent in monocular SLAM and ensures robust relocalization. Experiments demonstrate that our system effectively compensates for calibration errors, achieving an average absolute trajectory error of 1.09 mm. The reconstructed scenes reach an average target registration error of 2.16 mm, with high visual fidelity (SSIM = 0.69, PSNR = 22.19). These results indicate that our system provides a practical and cost-efficient solution for arthroscopic navigation, bridging the gap between optical tracking and purely vision-based systems, and paving the way toward clinically deployable, fully vision-based arthroscopic guidance.

eess.IV↗

arXiv2025

Constrained Natural Language Action Planning for Resilient Embodied Systems

Replicating human-level intelligence in the execution of embodied tasks remains challenging due to the unconstrained nature of real-world environments. Novel use of large language models (LLMs) for task planning seeks to address the previously intractable state/action space of complex planning tasks, but hallucinations limit their reliability, and thus, viability beyond a research context. Additionally, the prompt engineering required to achieve adequate system performance lacks transparency, and thus, repeatability. In contrast to LLM planning, symbolic planning methods offer strong reliability and repeatability guarantees, but struggle to scale to the complexity and ambiguity of real-world tasks. We introduce a new robotic planning method that augments LLM planners with symbolic planning oversight to improve reliability and repeatability, and provide a transparent approach to defining hard constraints with considerably stronger clarity than traditional prompt engineering. Importantly, these augmentations preserve the reasoning capabilities of LLMs and retain impressive generalization in open-world environments. We demonstrate our approach in simulated and real-world environments. On the ALFWorld planning benchmark, our approach outperforms current state-of-the-art methods, achieving a near-perfect 99% success rate. Deployment of our method to a real-world quadruped robot resulted in 100% task success compared to 50% and 30% for pure LLM and symbolic planners across embodied pick and place tasks. Our approach presents an effective strategy to enhance the reliability, repeatability and transparency of LLM-based robot planners while retaining their key strengths: flexibility and generalizability to complex real-world environments. We hope that this work will contribute to the broad goal of building resilient embodied intelligent systems.

cs.RO↗

arXiv2025

Egosurg: Arbitrary view synthesis for egocentric replay of operating room workflows from ambient cameras

Observing surgical practice has historically relied on fixed vantage points or recollections, leaving the egocentric perspectives that shape clinical decisions undocumented. Ambient fixed cameras capture the operating room (OR) at room scale but cannot recover what any individual team member actually saw. We present EgoSurg, a framework that reconstructs dynamic OR scenes from sparse wall-mounted stereo video and renders arbitrary, role-specific egocentric views without instrumenting personnel or interfering with clinical workflow. EgoSurg initializes a per-timestamp 3D Gaussian Splatting representation from scale-aware stereo depth and refines it with an image-conditioned diffusion model that corrects auxiliary rendered views, mitigating artifacts caused by limited camera coverage, crowding, and occlusion. We evaluated the framework on four real robotic pulmonology procedures and two simulated full-workflow sessions across two hospital sites. Near-field reconstruction fidelity was consistent (PSNR 26.8~dB, SSIM .895) across five workflow phases and both sites, and synthesized egocentric views reached a PSNR of 17.8~dB and an SSIM of .766 against paired hand-held point-of-view recordings. We further demonstrate three case studies for intended use: adjudicating a simulated sterile field violation, replaying a procedure from role-specific viewpoints, and testing a counterfactual change in personnel position. These results indicate that existing ambient camera infrastructure can be turned into a navigable 3D record of surgical work, supporting retrospective review of safety events, training, and workflow analysis from every angle.

cs.CV↗

arXiv2025

Explainable AI for Automated User-specific Feedback in Surgical Skill Acquisition

Traditional surgical skill acquisition relies heavily on expert feedback, yet direct access is limited by faculty availability and variability in subjective assessments. While trainees can practice independently, the lack of personalized, objective, and quantitative feedback reduces the effectiveness of self-directed learning. Recent advances in computer vision and machine learning have enabled automated surgical skill assessment, demonstrating the feasibility of automatic competency evaluation. However, it is unclear whether such Artificial Intelligence (AI)-driven feedback can contribute to skill acquisition. Here, we examine the effectiveness of explainable AI (XAI)-generated feedback in surgical training through a human-AI study. We create a simulation-based training framework that utilizes XAI to analyze videos and extract surgical skill proxies related to primitive actions. Our intervention provides automated, user-specific feedback by comparing trainee performance to expert benchmarks and highlighting deviations from optimal execution through understandable proxies for actionable guidance. In a prospective user study with medical students, we compare the impact of XAI-guided feedback against traditional video-based coaching on task outcomes, cognitive load, and trainees' perceptions of AI-assisted learning. Results showed improved cognitive load and confidence post-intervention. While no differences emerged between the two feedback types in reducing performance gaps or practice adjustments, trends in the XAI group revealed desirable effects where participants more closely mimicked expert practice. This work encourages the study of explainable AI in surgical education and the development of data-driven, adaptive feedback mechanisms that could transform learning experiences and competency assessment.

cs.HC↗

arXiv2025

Beyond Rigid AI: Towards Natural Human-Machine Symbiosis for Interoperative Surgical Assistance

Emerging surgical data science and robotics solutions, especially those designed to provide assistance in situ, require natural human-machine interfaces to fully unlock their potential in providing adaptive and intuitive aid. Contemporary AI-driven solutions remain inherently rigid, offering limited flexibility and restricting natural human-machine interaction in dynamic surgical environments. These solutions rely heavily on extensive task-specific pre-training, fixed object categories, and explicit manual-prompting. This work introduces a novel Perception Agent that leverages speech-integrated prompt-engineered large language models (LLMs), segment anything model (SAM), and any-point tracking foundation models to enable a more natural human-machine interaction in real-time intraoperative surgical assistance. Incorporating a memory repository and two novel mechanisms for segmenting unseen elements, Perception Agent offers the flexibility to segment both known and unseen elements in the surgical scene through intuitive interaction. Incorporating the ability to memorize novel elements for use in future surgeries, this work takes a marked step towards human-machine symbiosis in surgical procedures. Through quantitative analysis on a public dataset, we show that the performance of our agent is on par with considerably more labor-intensive manual-prompting strategies. Qualitatively, we show the flexibility of our agent in segmenting novel elements (instruments, phantom grafts, and gauze) in a custom-curated dataset. By offering natural human-machine interaction and overcoming rigidity, our Perception Agent potentially brings AI-based real-time assistance in dynamic surgical environments closer to reality.

cs.RO↗

arXiv2025

Privacy-Preserving Operating Room Workflow Analysis using Digital Twins

The operating room (OR) is a complex environment where optimizing workflows is critical to reduce costs and improve patient outcomes. While computer vision approaches for automatic recognition of perioperative events can identify bottlenecks for OR optimization, privacy concerns limit the use of OR videos for automated event detection. We propose a two-stage pipeline for privacy-preserving OR video analysis and event detection. First, we leverage vision foundation models for depth estimation and semantic segmentation to generate de-identified Digital Twins (DT) of the OR from conventional RGB videos. Second, we employ the SafeOR model, a fused two-stream approach that processes segmentation masks and depth maps for OR event detection. Evaluation on an internal dataset of 38 simulated surgical trials with five event classes shows that our DT-based approach achieves performance on par with -- and sometimes better than -- raw RGB video-based models for OR event detection. Digital Twins enable privacy-preserving OR workflow analysis, facilitating the sharing of de-identified data across institutions and potentially enhancing model generalizability by mitigating domain-specific appearance differences.

cs.CV↗

arXiv2025

Position: Foundation Models Need Digital Twin Representations

Current foundation models (FMs) rely on token representations that directly fragment continuous real-world multimodal data into discrete tokens. They limit FMs to learning real-world knowledge and relationships purely through statistical correlation rather than leveraging explicit domain knowledge. Consequently, current FMs struggle with maintaining semantic coherence across modalities, capturing fine-grained spatial-temporal dynamics, and performing causal reasoning. These limitations cannot be overcome by simply scaling up model size or expanding datasets. This position paper argues that the machine learning community should consider digital twin (DT) representations, which are outcome-driven digital representations that serve as building blocks for creating virtual replicas of physical processes, as an alternative to the token representation for building FMs. Finally, we discuss how DT representations can address these challenges by providing physically grounded representations that explicitly encode domain knowledge and preserve the continuous nature of real-world processes.

cs.LG↗

arXiv2025

Online Reasoning Video Segmentation with Just-in-Time Digital Twins

Reasoning segmentation (RS) aims to identify and segment objects of interest based on implicit text queries. As such, RS is a catalyst for embodied AI agents, enabling them to interpret high-level commands without requiring explicit step-by-step guidance. However, current RS approaches rely heavily on the visual perception capabilities of multimodal large language models (LLMs), leading to several major limitations. First, they struggle with queries that require multiple steps of reasoning or those that involve complex spatial/temporal relationships. Second, they necessitate LLM fine-tuning, which may require frequent updates to maintain compatibility with contemporary LLMs and may increase risks of catastrophic forgetting during fine-tuning. Finally, being primarily designed for static images or offline video processing, they scale poorly to online video data. To address these limitations, we propose an agent framework that disentangles perception and reasoning for online video RS without LLM fine-tuning. Our innovation is the introduction of a just-in-time digital twin concept, where -- given an implicit query -- a LLM plans the construction of a low-level scene representation from high-level video using specialist vision models. We refer to this approach to creating a digital twin as "just-in-time" because the LLM planner will anticipate the need for specific information and only request this limited subset instead of always evaluating every specialist model. The LLM then performs reasoning on this digital twin representation to identify target objects. To evaluate our approach, we introduce a new comprehensive video reasoning segmentation benchmark comprising 200 videos with 895 implicit text queries. The benchmark spans three reasoning categories (semantic, spatial, and temporal) with three different reasoning chain complexity.

cs.CV↗