Classification driven adaptive sensing improves measurement efficiency in image tasks
Classification-oriented adaptive sensing via posterior sampling
Computer Vision and Pattern Recognition
Summary
Getting the right data efficiently is important for making smart decisions from images. The authors studied how to choose what parts of an image to measure next, focusing on helping a computer classify the image rather than just reconstruct it perfectly. They used a new method that looks at uncertainty related to different categories, improving classification accuracy while reducing the amount of data needed. Their tests on popular image datasets show this method can save measurements with a small cost to how well the image itself can be rebuilt.
What this means in practice
- •For image processing engineers: Select minimal image data to classify objects faster with controlled quality trade-offs using the proposed posterior uncertainty criterion.
- •For medical imaging teams: Reduce measurement cost in imaging systems by prioritizing data acquisition that improves diagnosis-related classification rather than full image reconstruction.
Authors
Andriy Enttsel, Maxime Rousselot, Vincent Corlay
Abstract
Recent advances in diffusion models have enabled high-performance, instance-adaptive compressed sensing through posterior sampling, without task-specific policy training. Existing methods select sensing probes by maximizing total posterior signal variance and are therefore primarily reconstruction-driven. We introduce a classification-driven extension motivated by the closed-form posterior covariance of a class-conditional Gaussian mixture model, which decomposes into within-class and between-class uncertainty. Using calibrated soft classifier outputs, we estimate these uncertainty terms from diffusion posterior samples and propose a classification-oriented criterion for selecting the dominant sensing direction in the unmeasured subspace. Experiments on MNIST and CIFAR-10 compare the resulting classification accuracy, measurement cost, and reconstruction quality with those of reconstruction-oriented counterparts. The results identify regimes in which semantic posterior uncertainty yields a more favorable classification--measurement trade-off and quantify the associated reconstruction cost.