Papers for

computer vision system developers

Papers whose findings have a practical use for this group, as judged from the abstract. Open a paper to read what it means in practice.

When cross-view consistency matters for robot vision control tasks

When Does Cross-View Consistency Matter in Multi-View Appearance Augmentation? A Controlled Study for Visuomotor Policies

Abstract: Multi-view visual augmentation often seeks consistency across cameras, but it is unclear which forms of consistency matter for downstream control. We use simulation-state replay to manipulate cross-view coupling while preserving the per-camera perturbation distributions, physical trajectories, and action labels. Across seven LIBERO manipulation tasks, marginal-matched changes in global appearance, ghost injection, and target color have small average costs for a late-fusion Diffusion Policy: point estimates are near zero or negative, and the largest upper endpoint of the 90% posterior credible intervals is 2.55 percentage points. In contrast, geometrically anti-coupled observations reduce success by 3.30 percentage points relative to coupled observations (90% CrI [0.64, 6.28]), with displacement magnitude, direction marginals, and the number of affected views matched. Camera interventions reveal substantial dependence on wrist-view geometry, while feature analysis shows appearance sensitivity patterns associated with the factors randomized during training. ACT shows a higher target-color cost estimate than DP, with uncertainty that includes zero and sensitivity to task composition. Within the tested tasks and policy configurations, these findings motivate selective consistency constraints: appearance coupling can sometimes be relaxed, whereas task-relevant geometry and cross-view spatial correspondence warrant protection.

Fri 13 NovRoboticsComputer Vision and Pattern Recognition
The gist
Keeping images from multiple cameras consistent is common in robotic vision, but it’s unclear which types of consistency help robots perform better. The authors studied how different changes in camera views affect robot control in simulated tasks, finding that appearance changes usually have little effect, while inconsistencies in geometry between views hurt success more. They highlight that preserving geometric alignment and spatial correspondence between camera views is more important than keeping appearance exactly the same. This suggests robot vision systems can sometimes simplify appearance constraints without losing performance.
Open → 2611.17751v1

Multimodal models improve fine perception with blinking teacher method

Teach with Blinks: Breaking Linguistic Momentum in On-Policy Distillation for Fine-Grained Multimodal Perception

Abstract: On-policy distillation (OPD) has emerged as an effective post-training approach to improving fine-grained perception in multimodal large language models by amortizing costly inference-time zooming operations into a single-pass recipe. However, vanilla OPD is inherently susceptible to linguistic momentum: the teacher can follow the student's erroneous prefixes, thus weakening corrective supervision. To address this, we propose Blink On-Policy Distillation (BlinkOPD), in which the teacher "blinks" at every token: it reads the same prefix once with its eyes open and once with its eyes closed. A first-glance gain further strengthens the correction on rollouts that conflict with the teacher. By extracting more effective supervision from each on-policy trajectory, BlinkOPD achieves state-of-the-art average performance across six fine-grained benchmarks using only 500 training images (12x less data) than existing OPD methods. Beyond fine-grained perception, BlinkOPD also demonstrates strong zero-shot improvements across ten held-out benchmarks spanning general visual perception, visual hallucination, and multimodal reasoning abilities. Code: https://github.com/Supltz/BlinkOPD

Thu 12 NovComputer Vision and Pattern Recognition
The gist
Fine-grained recognition in AI models that understand both images and language can be slow and error-prone. The paper finds that when teaching a smaller model to imitate a larger 'teacher' model's understanding, the teacher might get misled by the student's mistakes. To fix this, the authors introduce a method where the teacher 'blinks'—sometimes paying full attention, sometimes ignoring parts of what the student predicts—to provide better guidance. This approach yields better accuracy using less training data and also improves performance on tasks the model wasn't specifically trained for.
Open → 2611.16402v1