HoughNet: Integrating Near and Long-Range Evidence for Visual Detection (2021-04-14T00:00:00.000000Z)

TL;DR

The effectiveness of HoughNet is validated 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 the voting module consistently improves performance in all cases.

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 <inline-formula><tex-math notation="LaTeX">$AP$</tex-math><alternatives><mml:math><mml:mrow><mml:mi>A</mml:mi><mml:mi>P</mml:mi></mml:mrow></mml:math><inline-graphic xlink:href="samet-ieq1-3200413.gif"/></alternatives></inline-formula> (and 65.1 <inline-formula><tex-math notation="LaTeX">$AP_{50}$</tex-math><alternatives><mml:math><mml:mrow><mml:mi>A</mml:mi><mml:msub><mml:mi>P</mml:mi><mml:mn>50</mml:mn></mml:msub></mml:mrow></mml:math><inline-graphic xlink:href="samet-ieq2-3200413.gif"/></alternatives></inline-formula>), 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.

Authors

Nermin Samet

2 papers

Emre Akbas

3 papers

Samet Hicsonmez

1 papers

TL;DR

Abstract

Authors

References78 items

One Metric to Measure Them All: Localisation Recall Precision (LRP) for Evaluating Visual Detection Tasks

RelationNet++: Bridging Visual Representations for Object Detection via Transformer Decoder

BorderDet: Border Feature for Dense Object Detection

HoughNet: Integrating near and long-range evidence for bottom-up object detection

End-to-End Object Detection with Transformers

Exploring Self-Attention for Image Recognition

Memory Enhanced Global-Local Aggregation for Video Object Detection

BlendMask: Top-Down Meets Bottom-Up for Instance Segmentation

PyTorch: An Imperative Style, High-Performance Deep Learning Library

FreeAnchor: Learning to Match Anchors for Visual Object Detection

HoughNet: Neural Network Architecture for Vanishing Points Detection

Object Guided External Memory Network for Video Object Detection

MMDetection: Open MMLab Detection Toolbox and Benchmark

LVIS: A Dataset for Large Vocabulary Instance Segmentation

Learning Temporal Pose Estimation from Sparsely-Labeled Videos

Deep Hough Voting for 3D Object Detection in Point Clouds

CenterNet: Keypoint Triplets for Object Detection

Objects as Points

Libra R-CNN: Towards Balanced Learning for Object Detection

FCOS: Fully Convolutional One-Stage Object Detection

Progressive Sparse Local Attention for Video Object Detection

Feature Selective Anchor-Free Module for Single-Shot Object Detection

Bottom-Up Object Detection by Grouping Extreme and Center Points

Region Proposal by Guided Anchoring

CornerNet: Detecting Objects as Paired Keypoints

Localization Recall Precision (LRP): A New Performance Metric for Object Detection

Analysis of the discriminative generalized Hough transform as a proposal generator for a deep network in automatic pedestrian and car detection

Simple Baselines for Human Pose Estimation and Tracking

YOLOv3: An Incremental Improvement

PersonLab: Person Pose Estimation and Instance Segmentation with a Bottom-Up, Part-Based, Geometric Embedding Model

Path Aggregation Network for Instance Segmentation

Cascade R-CNN: Delving Into High Quality Object Detection

Relation Networks for Object Detection

Non-local Neural Networks

Single-Shot Refinement Neural Network for Object Detection

CoupleNet: Coupling Global Structure with Local Parts for Object Detection

Focal Loss for Dense Object Detection

Deep Layer Aggregation

Attention is All you Need

Accurate Single Stage Detector Using Recurrent Rolling Convolution

Soft-NMS — Improving Object Detection with One Line of Code

Unpaired Image-to-Image Translation Using Cycle-Consistent Adversarial Networks

Flow-Guided Feature Aggregation for Video Object Detection

Mask R-CNN

DSSD : Deconvolutional Single Shot Detector

YOLO9000: Better, Faster, Stronger

Feature Pyramid Networks for Object Detection

Realtime Multi-person 2D Pose Estimation Using Part Affinity Fields

Image-to-Image Translation with Conditional Adversarial Networks

Associative Embedding: End-to-End Learning for Joint Detection and Grouping

R-FCN: Object Detection via Region-based Fully Convolutional Networks

Subcategory-Aware Convolutional Neural Networks for Object Proposals and Detection

The Cityscapes Dataset for Semantic Urban Scene Understanding

Human Pose Estimation Using Deep Consensus Voting

Deep Residual Learning for Image Recognition

SSD: Single Shot MultiBox Detector

Multi-Scale Context Aggregation by Dilated Convolutions

Faster R-CNN: Towards Real-Time Object Detection with Region Proposal Networks

Adam: A Method for Stochastic Optimization

Fully convolutional networks for semantic segmentation

ImageNet Large Scale Visual Recognition Challenge

Object detection through search with a foveated visual system

Microsoft COCO: Common Objects in Context

Latent Hough Transform for Object Detection

Are we ready for autonomous driving? The KITTI vision benchmark suite

Object Detection with Discriminatively Trained Part Based Models

On Detection of Multiple Object Instances Using Hough Transforms

A review of log-polar imaging for visual perception in robotics

Looking and Acting: Vision and eye movements in natural behaviour

Discriminative generalized hough transform for object dectection

Object detection using a max-margin Hough transform

Class-specific Hough forests for object detection

Robust Object Detection with Interleaved Categorization and Segmentation

Shape Matching and Object Recognition Using Shape Contexts

Machine Analysis of Bubble Chamber Pictures

degree from TOBB Economics and Technology University, Computer Engineering Department in 2013, and B.Sc. degree from Istanbul Technical University, Electronics Engineering