Adaptive Observation Control for Active Visual Recognition

Key points

• Active vision can adapt by observing a coarse scene and then selecting informative fixations, but the public AdaptiveNN-DeiT-S pipeline still fixes both field of view and observation count.1
• We extend this fixed-step recognizer with a controllable observation interface that adapts both the field of view and the number of fixations to the input.1
• Experiments across fine-grained benchmarks, full ImageNet-1K, and ImageNet-A/R-derived classification splits show that the interface remains competitive with fixed-step references while exposing a controllable observation budget.1
• On sequential ImageNet-1K, the learned stop rule outperforms fixed-step truncation of the same checkpoint, sustaining 81.9% top-1 using 2.45 of the four focus steps and exceeding three-step truncation at lower cost.1

Sources (1)

• [1]Adaptive Observation Control for Active Visual Recognition
  arXiv (AI, ML, NLP, CV, robotics, multi-agent) · Oct 14, 05:45 AM

  Visual recognition models usually process a fixed view under a fixed computation budget, although images often benefit from different scales and amounts of spatial evidence.

  Active vision can adapt by observing a coarse scene and then selecting informative fixations, but the public AdaptiveNN-DeiT-S pipeline still fixes both field of view and observation count.

Extractive summary: sentences quoted from the sources.