HyperNet: Towards Accurate Region Proposal Generation and Joint Object Detection

Abstract

Almost all of the current top-performing object detection networks employ region proposals to guide the search for object instances. State-of-the-art region proposal methods usually need several thousand proposals to get high recall, thus hurting the detection efficiency. Although the latest Region Proposal Network method gets promising detection accuracy with several hundred proposals, it still struggles in small-size object detection and precise localization (e.g., large IoU thresholds), mainly due to the coarseness of its feature maps. In this paper, we present a deep hierarchical network, namely HyperNet, for handling region proposal generation and object detection jointly. Our HyperNet is primarily based on an elaborately designed Hyper Feature which aggregates hierarchical feature maps first and then compresses them into a uniform space. The Hyper Features well incorporate deep but highly semantic, intermediate but really complementary, and shallow but naturally high-resolution features of the image, thus enabling us to construct HyperNet by sharing them both in generating proposals and detecting objects via an end-to-end joint training strategy. For the deep VGG16 model, our method achieves completely leading recall and state-of-the-art object detection accuracy on PASCAL VOC 2007 and 2012 using only 100 proposals per image. It runs with a speed of 5 fps (including all steps) on a GPU, thus having the potential for real-time processing.

Accepted as a spotlight oral paper by CVPR 2016

Cited by in corpus (34)

• DSSD : Deconvolutional Single Shot Detector
• Feature Pyramid Networks for Object Detection
• Evolution of Image Segmentation using Deep Convolutional Neural Network: A Survey
• Deep Learning for Generic Object Detection: A Survey
• Single-Shot Refinement Neural Network for Object Detection
• RON: Reverse Connection with Objectness Prior Networks for Object Detection
• Selective Feature Connection Mechanism: Concatenating Multi-layer CNN Features with a Feature Selector
• MDSSD: Multi-scale Deconvolutional Single Shot Detector for Small Objects
• A Comparison of CNN-based Face and Head Detectors for Real-Time Video Surveillance Applications
• WordSup: Exploiting Word Annotations for Character based Text Detection
• Improving Object Detection from Scratch via Gated Feature Reuse
• Zoom Out-and-In Network with Recursive Training for Object Proposal
• Feature Agglomeration Networks for Single Stage Face Detection
• Structure Inference Net: Object Detection Using Scene-Level Context and Instance-Level Relationships
• Residual Features and Unified Prediction Network for Single Stage Detection
• On The Stability of Video Detection and Tracking
• Cross-X Learning for Fine-Grained Visual Categorization
• Feature Enhancement Network: A Refined Scene Text Detector
• Multi-scale Location-aware Kernel Representation for Object Detection
• Single-Shot Bidirectional Pyramid Networks for High-Quality Object Detection
• A Survey on Deep Domain Adaptation and Tiny Object Detection Challenges, Techniques and Datasets
• ScratchDet: Training Single-Shot Object Detectors from Scratch
• Weaving Multi-scale Context for Single Shot Detector
• IPG-Net: Image Pyramid Guidance Network for Small Object Detection
• Pruning Deep Neural Networks using Partial Least Squares
• Zoom Out-and-In Network with Map Attention Decision for Region Proposal and Object Detection
• Object Detection based on Region Decomposition and Assembly
• Deep Feature Pyramid Reconfiguration for Object Detection
• Detecting Heads using Feature Refine Net and Cascaded Multi-Scale Architecture
• A Taught-Obesrve-Ask (TOA) Method for Object Detection with Critical Supervision
• DSNet: Deep and Shallow Feature Learning for Efficient Visual Tracking
• Improved detection of small objects in road network sequences
• Feature Selective Networks for Object Detection
• Precise Box Score: Extract More Information from Datasets to Improve the Performance of Face Detection