Probabilistic Anchor Assignment with IoU Prediction for Object Detection

By Kang-jik Kim, Hee Seok Lee. European Conference on Computer Vision, 2020

501 citations recorded by the source corpus.

In object detection, determining which anchors to assign as positive or negative samples, known as anchor assignment, has been revealed as a core procedure that can significantly affect a model's performance. In this paper we propose a novel anchor assignment strategy that adaptively separates anchors into positive and negative samples for a ground truth bounding box according to the model's learning status such that it is able to reason about the separation in a probabilistic manner. To do so we first calculate the scores of anchors conditioned on the model and fit a probability distribution to these scores. The model is then trained with anchors separated into positive and negative samples according to their probabilities. Moreover, we investigate the gap between the training and testing objectives and propose to predict the Intersection-over-Unions of detected boxes as a measure of localization quality to reduce the discrepancy. The combined score of classification and localization qualities serving as a box selection metric in non-maximum suppression well aligns with the proposed anchor assignment strategy and leads significant performance improvements. The proposed methods only add a single convolutional layer to RetinaNet baseline and does not require multiple anchors per location, so are efficient. Experimental results verify the effectiveness of the proposed methods. Especially, our models set new records for single-stage detectors on MS COCO test-dev dataset with various backbones. Code is available at this https URL.

Subfields: Computer Vision and Pattern Recognition

View Paper: https://www.semanticscholar.org/paper/92ad35b9e4b0ab761bcb1649fb8f5812f831052b

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

Focal and Efficient IOU Loss for Accurate Bounding Box Regression (2021, methodology): “All other hyper-parameters follow the settings in ATSS [11] if not specifically noted.”
DETRs with Collaborative Hybrid Assignments Training (2022, background): “PAA [17] adaptively separates anchors into positive and negative samples in a probabilistic manner.”
Group Fisher Pruning for Practical Network Compression (2021, methodology): “We prune one-stage meth-Group ods including RetinaNet (Lin et al., 2017b), FSAF (Zhu et al., 2019), ATSS (Zhang et al., 2020) and PAA (Kim & Lee, 2020), and two-stage method Faster R-CNN (Ren et al., 2015) to extensively validate the effectiveness of our method.”
Probabilistic two-stage detection (2021, background): “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.”
Generalized Focal Loss V2: Learning Reliable Localization Quality Estimation for Dense Object Detection (2020, methodology): “Following the similar paradigm, IoU-Net [13], Fitness NMS [34], MS R-CNN [12], IoUaware [36], PAA [14] utilize a separate branch to perform LQE in the IoU form.”
TOOD: Task-aligned One-stage Object Detection (2021, methodology): “Different from the previous works using a centerness branch [27] or an IoU branch [9] which can only adjust the classification prediction based on either classification features or localization features, we align the two predictions by considering both tasks jointly using the computed task-interacti”
OTA: Optimal Transport Assignment for Object Detection (2021, background): “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.”
Oriented RepPoints for Aerial Object Detection (2021, methodology): “Methods Max-IoU [28] ATSS [50] PAA [13] CFA [7] APAA”
Towards Large-Scale Small Object Detection: Survey and Benchmarks (2022, background): “1) Sample-Oriented Methods: One of the most critical procedures of training a learning-based detector is the sampling (often coexists with assignment), which has led to signiﬁcant progress in generic object detection [53], [54].”
RTMDet: An Empirical Study of Designing Real-Time Object Detectors (2022, background): “Later practices [37, 70, 95, 98] further explore different matching criteria such as object centers [70, 98].”
YOLOv7: Trainable Bag-of-Freebies Sets New State-of-the-Art for Real-Time Object Detectors (2022, methodology): “However, in recent years, if we take object detection as an example, researchers often use the quality and distribution of prediction output by the network, and then consider together with the ground truth to use some calculation and optimization methods to generate a reliable soft label [61, 8, 36,”
YOLOX: Exceeding YOLO Series in 2021 (2021, background): “SimOTA ﬁrst calculates pair-wise matching degree, represented by cost [4, 5, 12, 2] or quality [33] for each prediction-gt pair.”
Detecting tiny objects in aerial images: A normalized Wasserstein distance and a new benchmark (2022, unknown): “Probabilistic Anchor Assignment (PAA) (Kim and Lee, 2020) assumes that the distribution of joint loss for pos/neg samples follows the Gaussian distribution.”
A Normalized Gaussian Wasserstein Distance for Tiny Object Detection (2021, unknown): “[9] introduce Probabilistic Anchor Assignment (PAA) by assuming that the distribution of joint loss for pos/neg samples follows the Gaussian distribution.”