Om Rastogi

A video large language model can watch two clips of the same person in the same room and name every object in both, yet fail to tell you which one moves faster, which way a hand travels, or how hard a door is shut. That gap matters because Video-LLMs are increasingly the perceptual front end of world models: auto-labeling state transitions, acting as reward and success detectors for model-based RL, grounding instructions in what an agent sees. Each of those roles presupposes a competence at reading motion that, we show, these models do not have. MotionBlind is a contrastive, minimal-pairs benchmark of self-recorded clip pairs that are near-identical except in motion, each paired with two complementary yes/no questions, so a model must answer all four items correctly (Instance Accuracy, I_Acc) to score the instance. Single-frame, appearance and language-only shortcuts all collapse toward the 6.25% chance floor. It complements TimeBlind with physically grounded categories (speed, magnitude, direction) that are hard to source and label from uncontrolled internet video but are precisely the variables a world model must predict. Across two suites we run a controlled study of six open Video-LLMs and two frontier models over frame budgets 1-24 and four frame-selection strategies. Open models sit near the chance floor; a larger model does no better; and neither more frames nor dynamic selection, which changes which frames are seen and not whether motion is read, closes the gap. The task genuinely needs video in the right order: removing it zeroes every model, and shuffling frames collapses accuracy to chance. Only Gemini 3.1 Pro clears the benchmark overall, yet it too fails on the purely rate-defined categories.

AdShot is a benchmark for evaluating multimodal large language models on advertisement video clipping, the task of shortening a 30-second ad to a 15-second cut while preserving brand messaging and narrative coherence. It comprises 823 video pairs across 194 brands sourced from real advertisement archives, with an automated evaluation harness scoring temporal coherence, brand retention, and narrative quality across models including Qwen-Omni, MiniCPM-o, and Aria.

The paper is focused on providing a color thresholding method, that is independent of lighting conditions in a video. A two-layered process is proposed, a luminosity filter followed by a variable color threshold. The approach is non-parametric and deterministic. For applications like Lane Detection System, there can be multiple lighting conditions depending on the weather conditions and time of day. Being able to use a color mask, instead of edging techniques for road markers would be robust.