DAVE -- A Detect-and-Verify Paradigm for Low-Shot Counting

TL;DR

DAVE introduces a detect-and-verify approach for low-shot counting that improves accuracy by generating high-recall detections and verifying them to reduce false positives, outperforming existing methods in count accuracy and detection quality.

Contribution

The paper presents DAVE, a novel low-shot counting method that combines detection and verification to enhance count accuracy and detection quality, addressing limitations of previous density and detection-based approaches.

Findings

DAVE outperforms top density-based counters by ~20% in MAE.

DAVE surpasses recent detection-based counters by ~20% in detection quality.

Sets new state-of-the-art in zero-shot and text-prompt-based counting.

Abstract

Low-shot counters estimate the number of objects corresponding to a selected category, based on only few or no exemplars annotated in the image. The current state-of-the-art estimates the total counts as the sum over the object location density map, but does not provide individual object locations and sizes, which are crucial for many applications. This is addressed by detection-based counters, which, however fall behind in the total count accuracy. Furthermore, both approaches tend to overestimate the counts in the presence of other object classes due to many false positives. We propose DAVE, a low-shot counter based on a detect-and-verify paradigm, that avoids the aforementioned issues by first generating a high-recall detection set and then verifying the detections to identify and remove the outliers. This jointly increases the recall and precision, leading to accurate counts. DAVE outperforms the top density-based counters by ~20% in the total count MAE, it outperforms the most recent detection-based counter by ~20% in detection quality and sets a new state-of-the-art in zero-shot as well as text-prompt-based counting.