HoughNet: Integrating near and long-range evidence for visual detection

TL;DR

HoughNet introduces a voting-based, anchor-free object detection method that combines near and long-range evidence, achieving competitive results across multiple visual recognition tasks by leveraging a generalized Hough Transform-inspired mechanism.

Contribution

The paper proposes HoughNet, a novel detection approach that integrates near and long-range evidence through a voting mechanism, enhancing detection accuracy over existing methods.

Findings

Achieves 46.4 AP on COCO dataset, competitive with state-of-the-art.

Improves performance in video detection, segmentation, 3D detection, and pose estimation.

Voting mechanism consistently boosts results across various tasks.

Abstract

This paper presents HoughNet, a one-stage, anchor-free, voting-based, bottom-up object detection method. Inspired by the Generalized Hough Transform, HoughNet determines the presence of an object at a certain location by the sum of the votes cast on that location. Votes are collected from both near and long-distance locations based on a log-polar vote field. Thanks to this voting mechanism, HoughNet is able to integrate both near and long-range, class-conditional evidence for visual recognition, thereby generalizing and enhancing current object detection methodology, which typically relies on only local evidence. On the COCO dataset, HoughNet's best model achieves $46.4$ $AP$ (and $65.1$ $AP_{50}$), performing on par with the state-of-the-art in bottom-up object detection and outperforming most major one-stage and two-stage methods. We further validate the effectiveness of our proposal in other visual detection tasks, namely, video object detection, instance segmentation, 3D object detection and keypoint detection for human pose estimation, and an additional "labels to photo" image generation task, where the integration of our voting module consistently improves performance in all cases. Code is available at https://github.com/nerminsamet/houghnet.