Receptive Field Block Net for Accurate and Fast Object Detection

By Songtao Liu, Di Huang, Yunhong Wang. 2017

1,541 citations recorded by the source corpus.

Current top-performing object detectors depend on deep CNN backbones, such as ResNet-101 and Inception, benefiting from their powerful feature representations but suffering from high computational costs. Conversely, some lightweight model based detectors fulfil real time processing, while their accuracies are often criticized. In this paper, we explore an alternative to build a fast and accurate detector by strengthening lightweight features using a hand-crafted mechanism. Inspired by the structure of Receptive Fields (RFs) in human visual systems, we propose a novel RF Block (RFB) module, which takes the relationship between the size and eccentricity of RFs into account, to enhance the feature discriminability and robustness. We further assemble RFB to the top of SSD, constructing the RFB Net detector. To evaluate its effectiveness, experiments are conducted on two major benchmarks and the results show that RFB Net is able to reach the performance of advanced very deep detectors while keeping the real-time speed. Code is available at this https URL.

Subfields: Computer Vision and Pattern Recognition

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Computer Vision and Pattern Recognition

Citations in the corpus

Bridging the Gap Between Anchor-Based and Anchor-Free Detection via Adaptive Training Sample Selection (2019, background): “Thereafter, plenty of works are presented to boost its performance in different aspects, such as fusing context information from different layers [24, 12, 69], training from scratch [50, 73], introducing new loss function [33, 6], anchor refinement and matching [66, 67], architecture redesign [21, 2”
YOLOv4: Optimal Speed and Accuracy of Object Detection (2020, background): “Common modules that can be used to enhance receptive field are SPP [25], ASPP [5], and RFB [47].”
Towards Accurate One-Stage Object Detection With AP-Loss (2019, background): “Table 3 lists the benchmark results comparing to recent state-of-the-art onestage detectors such as SSD [18], YOLOv2 [23], DSSD [6], DSOD [28], DES [36], RetinaNet [15], RefineDet [35], PFPNet [10], RFBNet [17].”
Small-Object Detection in Remote Sensing Images with End-to-End Edge-Enhanced GAN and Object Detector Network (2020, methodology): “[52] proposed a receptive field block (RFB) module that used the relationship between the size and eccentricity of receptive fields to enhance the feature discrimination and robustness.”
A Comparative Analysis of Object Detection Metrics with a Companion Open-Source Toolkit (2021, neutral): “…R-CNN [8] COCO AP@[.5:.05:.95]; AP@.50 R-CNN [52] PASCAL VOC 2007, 2010, 2012 AP; mAP (IOU=.50) RFB Net [78] PASCAL VOC 2007 mAP (IOU=.50) RFB Net [78] COCO AP@[.5:.05:.95]; AP@.50; AP@.75; AP S ; AP M ; AP L ReﬁneDet [79] PASCAL VOC 2007, 2012 mAP (IOU=.50) ReﬁneDet [79] COCO AP@[.5:.05:.95];…”
Multi-Scale Positive Sample Refinement for Few-Shot Object Detection (2020, methodology): “stage detectors use a single CNN to directly predict bounding boxes [29,30,25,24], and two-stage ones first generate region proposals and then classify them for decision making [12,11,31].”
CornerNet-Lite: Efficient Keypoint based Object Detection (2019, background): “RFBNet [33] proposes a new module which mimics the receptive ﬁeld of human vision systems to efﬁciently gather information across different scales.”
High-Resolution Representations for Labeling Pixels and Regions (2019, neutral): “RFBNet [64] VGG16 512 - 34.”
Designing Network Design Strategies Through Gradient Path Analysis (2022, background): “For example, use asymmetric computational units to extract features with different receptive ﬁelds [5, 1, 21]; (2) can automatically select suitable operation units with parameterized models for different inputs.”
A Comprehensive Review of YOLO: From YOLOv1 to YOLOv8 and Beyond (2023, methodology): “Examples of these methods are those for enlarging the receptive field [48, 51, 52], combining features [53, 49, 54, 55], and post-processing [56, 40, 57, 58] among others.”
Diversify and Match: A Domain Adaptive Representation Learning Paradigm for Object Detection (2019, background): “With the recent advances of the convolutional neural networks (CNNs), CNN-based methods [13, 12, 35, 30, 34, 26, 8, 46, 29] have achieved signiﬁcant progress in object detection based on ﬁne benchmarks [10, 27, 25].”
FFCA-YOLO for Small Object Detection in Remote Sensing Images (2024, positive): “2, which is inspired by RFB-s [52].”
Recent Advances in Deep Learning for Object Detection (2019, background): “proposed a Receptive Field Block Net (RFBNet) [108] to enhance the robustness and receptive fields via a receptive field block (RFB block).”
TPH-YOLOv5: Improved YOLOv5 Based on Transformer Prediction Head for Object Detection on Drone-captured Scenarios (2021, methodology): “There are also some additional blocks used in neck, like SPP [16], ASPP [5], RFB [31], CBAM [54].”