Scale-Aware Trident Networks for Object Detection

By Yanghao Li, Yuntao Chen, Naiyan Wang, Zhaoxiang Zhang. 2019

1,060 citations recorded by the source corpus.

Scale variation is one of the key challenges in object detection. In this work, we ﬁrst present a controlled experiment to investigate the effect of receptive ﬁelds for scale variation in object detection. Based on the ﬁndings from the exploration experiments, we propose a novel Trident Network (TridentNet) aiming to generate scale-speciﬁc feature maps with a uniform representational power. We construct a parallel multi-branch architecture in which each branch shares the same transformation parameters but with different receptive ﬁelds. Then, we adopt a scale-aware training scheme to specialize each branch by sampling object instances of proper scales for training. As a bonus, a fast approximation version of TridentNet could achieve signiﬁcant improvements without any additional parameters and computational cost compared with the vanilla detector. On the COCO dataset, our TridentNet with ResNet-101 backbone achieves state-of-the-art single-model results of 48.4 mAP. Codes are available at https://git.io/fj5vR.

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

Bridging the Gap Between Anchor-Based and Anchor-Free Detection via Adaptive Training Sample Selection (2019, methodology): “After that, lots of algorithms are proposed to improve its performance, including architecture redesign and reform [4, 9, 5, 28, 30], context and attention mechanism [2, 51, 38, 7, 44], multiscale training and testing [54, 41], training strategy and loss function [40, 52, 61, 17], feature fusion and”
Objects as Points (2019, background): “Sophisticated two-stage detectors [31,35,48,63] are more accurate, but also slower.”
Exploring Plain Vision Transformer Backbones for Object Detection (2022, background): “More recently, works including Trident Networks [33] and YOLOF [7] have revisited single-scale feature maps, but unlike our work they focus on a single-scale taken from a hierarchical backbone.”
ZeroWaste Dataset: Towards Deformable Object Segmentation in Cluttered Scenes (2021, methodology): “We perform fully-supervised object detection on ZeroWaste-f using the most widely used Mask R-CNN [39], RetinaNet [34], TridentNet [33] models, and we train DeepLabV3+ [14] as a semantic segmentation baseline.”
Generalized Focal Loss: Learning Qualified and Distributed Bounding Boxes for Dense Object Detection (2020, neutral): “6 ICCV19 TridentNet [16] R-101-DCN 36 X 1.”
CenterNet: Keypoint Triplets for Object Detection (2019, background): “, a [52, 22] focus on the architecture design, a [1, 10, 37, 47] focus on the contextual relationship, and a [23, 3] focus on the multi-scale unification.”
YOLOv4: Optimal Speed and Accuracy of Object Detection (2020, background): “Scale-Aware Trident Networks for Object Detection [42] TridentNet ResNet-101 - 2.”
Conditional DETR for Fast Training Convergence (2021, methodology): “Typical methods include YOLOv1 [30], CornerNet [21], ExtremeNet [50], CenterNet [49, 6], FCOS [39] and others [23, 28, 52, 19, 51, 22, 15, 46, 47].”
Learning Depth-Guided Convolutions for Monocular 3D Object Detection (2019, methodology): “Trident network [27] utilizes manually defined multi-head detectors for 2D detection.”
CentripetalNet: Pursuing High-Quality Keypoint Pairs for Object Detection (2020, neutral): “0 TridentNet(multi-scale) [18] ResNet-101-DCN 48.”
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.”
Rethinking Classification and Localization for Object Detection (2019, background): “Trident Network [24] generates scale-aware feature maps with multi-branch architecture.”
Injecting Semantic Concepts into End-to-End Image Captioning (2021, background): “Prevailing conventional models [25, 39, 79] which are based on an object detector to extract regional features.”
Imbalance Problems in Object Detection: A Review (2019, methodology): “• Efficient Featurized Image Pyramids [72] • Enriched Feature Guided Refinement Network [58] • Super Resolution for Small Objects [73] • Scale Aware Trident Network [74]”
Asymmetric Contextual Modulation for Infrared Small Target Detection (2020, background): “re-customizing the down-sampling scheme: Many studies emphasize that when designing CNNs, the receptive fields of predictors should match the object scale range [29, 20].”
Computation Reallocation for Object Detection (2019, methodology): “To reallocate the computation across spatial position, we utilize dilated convolution Li et al. (2019), Li et al. (2018).”
VisDrone-DET2019: The Vision Meets Drone Object Detection in Image Challenge Results (2019, background): “Scale-Aware Trident Networks for Object Detection (TridentNet) Xuzhang Zhang shantan@hust.edu.cn TridentNet [22] aims to generate scale-speciﬁc feature maps with a uniform representational power.”