Learning to Match Anchors for Visual Object Detection

Zhang, Xiaosong; Wan, Fang; Liu, Chang; Ji, Xiangyang; Ye, Qixiang

doi:10.1109/tpami.2021.3050494

Cited by 243 publications

(251 citation statements)

References 36 publications

Supporting

Mentioning

249

Contrasting

Unclassified

Order By: Relevance

“…Pang et al [14] proposed the Libra-RCNN, which aims at balanced learning for object detection. Several two-stage methods have achieved remarkable progress on accuracy [15,16] . One-stage detection methods benefit the speed of inference without resampling operations [17][18][19] .…”

Section: Related Workmentioning

confidence: 99%

“…Li [21] presented Gaussian proposal networks, which propose bounding ellipses as 2D Gaussian distributions on the image plane. Some detection methods that use keypoints for detection were proposed [16,22,23] . In addition, several methods were proposed to optimize the algorithms in bounding-box regression or some metrics when assigning labels [24][25][26] .…”

Section: Related Workmentioning

confidence: 99%

See 1 more Smart Citation

Two-stage lesion detection approach based on dimension-decomposition and 3D context

Jiao

Pan

Chen

et al. 2022

Tsinghua Sci. Technol.

View full text Add to dashboard Cite

Lesion detection in Computed Tomography (CT) images is a challenging task in the field of computer-aided diagnosis. An important issue is to locate the area of lesion accurately. As a branch of Convolutional Neural Networks (CNNs), 3D Context-Enhanced (3DCE) frameworks are designed to detect lesions on CT scans. The False Positives (FPs) detected in 3DCE frameworks are usually caused by inaccurate region proposals, which slow down the inference time. To solve the above problems, a new method is proposed, a dimension-decomposition region proposal network is integrated into 3DCE framework to improve the location accuracy in lesion detection.Without the restriction of "anchors" on ratios and scales, anchors are decomposed to independent "anchor strings".Anchor segments are dynamically combined in accordance with probability, and anchor strings with different lengths dynamically compose bounding boxes. Experiments show that the accurate region proposals generated by our model promote the sensitivity of FPs and spend less inference time compared with the current methods.

show abstract

Section: Related Workmentioning

confidence: 99%

Section: Related Workmentioning

confidence: 99%

Two-stage lesion detection approach based on dimension-decomposition and 3D context

Jiao

Pan

Chen

et al. 2022

Tsinghua Sci. Technol.

View full text Add to dashboard Cite

show abstract

“…That is, the positives and negatives are selected according to the preset IoU threshold [4]. Although some novel methods have been proposed to improve the label assignment strategy [37][38][39], these works do not take into account the characteristics of aerial image targets. Recently, some label assignment methods have been proposed for rotating aerial object detection [10,20,40].…”

Section: Object Detection In Aerial Imagesmentioning

confidence: 99%

“…It has been proved in some previous work that the detector can achieve good performance without using dense anchors during training [6,41,42]. For example, YOLOv3 [6] only uses one anchor with the highest IoU as the positive sample for training.…”

Section: Sparse Label Assignment For Efficient Training Sample Selectionmentioning

confidence: 99%

See 1 more Smart Citation

Sparse Label Assignment for Oriented Object Detection in Aerial Images

Miao

Zhou

et al. 2021

Remote Sensing

View full text Add to dashboard Cite

Object detection in aerial images has received extensive attention in recent years. The current mainstream anchor-based methods directly divide the training samples into positives and negatives according to the intersection-over-unit (IoU) of the preset anchors. This label assignment strategy assigns densely arranged samples for training, which leads to a suboptimal learning process and cause the model to suffer serious duplicate detections and missed detections. In this paper, we propose a sparse label assignment strategy (SLA) to select high-quality sparse anchors based on the posterior IoU of detections. In this way, the inconsistency between classification and regression is alleviated, and better performance can be achieved through balanced training. Next, to accurately detect small and densely arranged objects, we use a position-sensitive feature pyramid network (PS-FPN) with a coordinate attention module to extract position-sensitive features for accurate localization. Finally, the distance rotated IoU loss is proposed to eliminate the inconsistency between the training loss and the evaluation metric for better bounding box regression. Extensive experiments on the DOTA, HRSC2016, and UCAS-AOD datasets demonstrate the superiority of the proposed approach.

show abstract

Fracture R‐CNN: An anchor‐efficient anti‐interference framework for skull fracture detection in CT images

et al. 2022

View full text Add to dashboard Cite

Background Skull fracture, as a common traumatic brain injury, can lead to multiple complications including bleeding, leaking of cerebrospinal fluid, infection, and seizures. Automatic skull fracture detection (SFD) is of great importance, especially in emergency medicine. Purpose Existing algorithms for SFD, developed based on hand‐crafted features, suffer from low detection accuracy due to poor generalizability to unseen samples. Deploying deep detectors designed for natural images like Faster Region‐based Convolutional Neural Network (R‐CNN) for SFD can be helpful but are of high redundancy and with nonnegligible false detections due to the cranial suture and skull base interference. Therefore, we, for the first time, propose an anchor‐efficient anti‐interference deep learning framework named Fracture R‐CNN for accurate SFD with low computational cost. Methods The proposed Fracture R‐CNN is developed by incorporating the prior knowledge utilized in clinical diagnosis into the original Faster R‐CNN. Specifically, based on the distributions of skull fractures, we first propose an adaptive anchoring region proposal network (AA‐RPN) to generate proposals for diverse‐scale fractures with low computational complexity. Then, based on the prior knowledge that cranial sutures exist in the junctions of bones and usually contain sclerotic margins, we design an anti‐interference head (A‐Head) network to eliminate the cranial suture interference for better SFD detection. In addition, to further enhance the anti‐interference ability of the proposed A‐Head, a difficulty‐balanced weighted loss function is proposed to emphasize more on distinguishing the interference areas from the skull base and the cranial sutures during training. Results Experimental results demonstrate that the proposed Fracture R‐CNN outperforms the current state‐of‐the‐art (SOTA) deep detectors for SFD with a higher recall and fewer false detections. Compared to Faster R‐CNN, the proposed Fracture R‐CNN improves the average precision (AP) by 11.74% and the free‐response receiver operating characteristic (FROC) score by 11.08%. Through validating on various backbones, we further demonstrate the architecture independence of Fracture R‐CNN, making it extendable to other detection applications. Conclusions As the customized deep learning–based framework for SFD, Fracture R‐CNN can effectively overcome the unique challenges in SFD with less computational cost, leading to a better detection performance compared to the SOTA deep detectors. Moreover, we believe the prior knowledge explored for Fracture R‐CNN would shed new light on future deep learning approaches for SFD.

show abstract

Learning to Match Anchors for Visual Object Detection

Cited by 243 publications

References 36 publications

Two-stage lesion detection approach based on dimension-decomposition and 3D context

Two-stage lesion detection approach based on dimension-decomposition and 3D context

Sparse Label Assignment for Oriented Object Detection in Aerial Images

Fracture R‐CNN: An anchor‐efficient anti‐interference framework for skull fracture detection in CT images

Contact Info

Product

Resources

About