Learning to Match Anchors for Visual Object Detection

By Xiaosong Zhang, Fang Wan, Chang Liu, Xiangyang Ji, Qixiang Ye. 2021

235 citations recorded by the source corpus.

Modern CNN-based object detectors assign anchors for ground-truth objects under the restriction of object-anchor Intersection-over-Union (IoU). In this study, we propose a learning-to-match (LTM) method to break IoU restriction, allowing objects to match anchors in a flexible manner. LTM updates hand-crafted anchor assignment to “free” anchor matching by formulating detector training in the Maximum Likelihood Estimation (MLE) framework. During the training phase, LTM is implemented by converting the detection likelihood to anchor matching loss functions which are plug-and-play. Minimizing the matching loss functions drives learning and selecting features which best explain a class of objects with respect to both classification and localization. LTM is extended from anchor-based detectors to anchor-free detectors, validating the general applicability of learnable object-feature matching mechanism for visual object detection. Experiments on MS COCO dataset demonstrate that LTM detectors consistently outperform counterpart detectors with significant margins. The last but not the least, LTM requires negligible computational cost in both training and inference phases as it does not involve any additional architecture or parameter. Code has been made publicly available.

Subfields: Artificial Intelligence, 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 Artificial Intelligence

In Computer Vision and Pattern Recognition

Citations in the corpus

Generalized Focal Loss: Learning Qualified and Distributed Bounding Boxes for Dense Object Detection (2020, methodology): “Specifically, bounding box representation is modeled as a simple Dirac delta distribution [10, 18, 32, 26, 31], which is widely used over past years.”
A Survey of Visual Transformers (2021, methodology): “Such label assignment bears a resemblance to the method of learning to match anchors (AnchorFree) [119].”
Anti-aliasing Semantic Reconstruction for Few-Shot Semantic Segmentation (2021, background): “Over the past few years, we have witnessed the substantial progress of object detection and semantic segmentation [45, 46, 28, 48, 1, 14].”
Probabilistic two-stage detection (2021, methodology): “While one-stage detectors have achieved competitive performance (Zhang et al., 2020b; Kim & Lee, 2020; Zhang et al., 2019; Li et al., 2020b; Zhu et al., 2020a), they usually rely on heavier separate classiﬁcation and regression branches than two-stage models.”
Imbalance Problems in Object Detection: A Review (2019, methodology): “Compared to guided anchoring, free anchor [68] reports a lower relative improvement for average recall with 2.”
PP-YOLO: An Effective and Efficient Implementation of Object Detector (2020, methodology): “Anchor-based methods are still the mainstream of object detection [33, 21, 31, 32, 1, 22, 28, 9, 45, 2, 5, 37, 20, 4, 15], which evolved from early proposal based detectors, such as Fast R-CNN [11].”
Soft Anchor-Point Object Detection (2019, unknown): “4 FreeAnchor [32] R-101 9.”
OTA: Optimal Transport Assignment for Object Detection (2021, methodology): “Other recent advances [48, 19, 51, 16] suggest that the predicted confidence scores of each anchor could be a proper indicator to design dynamic assigning strategies, i.”
Rethinking Transformer-based Set Prediction for Object Detection (2020, methodology): “[39] formulates an MLE procedure to learn the matching between sliding windows and ground truth objects.”
Oriented RepPoints for Aerial Object Detection (2021, methodology): “Some recent samples assignment methods in general object detection, such as the ATSS [50], FreeAnchor [51], PAA [13] and OTA [6], employ a learning-tomatch optimization strategy [52] to choose the high-quality samples.”
Corner Proposal Network for Anchor-free, Two-stage Object Detection (2020, methodology): “FreeAnchor [47] used a larger resolution of input images (e.”
Generalized Focal Loss V2: Learning Reliable Localization Quality Estimation for Dense Object Detection (2020, unknown): “FreeAnchor [41] R-101 24 X 12.”