High-Level Semantic Feature Detection: A New Perspective for Pedestrian Detection

By Wei David Liu, Shengcai Liao, Weiqiang Ren, Weidong Hu, Yinan Yu. 2019

395 citations recorded by the source corpus.

Object detection generally requires sliding-window classifiers in tradition or anchor-based predictions in modern deep learning approaches. However, either of these approaches requires tedious configurations in windows or anchors. In this paper, taking pedestrian detection as an example, we provide a new perspective where detecting objects is motivated as a high-level semantic feature detection task. Like edges, corners, blobs and other feature detectors, the proposed detector scans for feature points all over the image, for which the convolution is naturally suited. However, unlike these traditional low-level features, the proposed detector goes for a higher-level abstraction, that is, we are looking for central points where there are pedestrians, and modern deep models are already capable of such a high-level semantic abstraction. Besides, like blob detection, we also predict the scales of the pedestrian points, which is also a straightforward convolution. Therefore, in this paper, pedestrian detection is simplified as a straightforward center and scale prediction task through convolutions. This way, the proposed method enjoys an anchor-free setting. Though structurally simple, it presents competitive accuracy and good speed on challenging pedestrian detection benchmarks, and hence leading to a new attractive pedestrian detector. Code and models will be available at https://github.com/liuwei16/CSP.

Subfields: Civil and Structural Engineering, 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 Civil and Structural Engineering

1. Ling Shao

In Computer Vision and Pattern Recognition

1. Luc Van Gool
2. Ming-Hsuan Yang
3. Xiaogang Wang
4. Dacheng Tao
5. Ling Shao
6. Yi Yang
7. Bernt Schiele
8. Andrew Zisserman

Citations in the corpus

• Bridging the Gap Between Anchor-Based and Anchor-Free Detection via Adaptive Training Sample Selection (2019, background): “CSP [37] only defines the center point of the object box as positives to detect pedestrians with fixed aspect ratio.”
• Global Pooling, More than Meets the Eye: Position Information is Encoded Channel-Wise in CNNs (2021, neutral): “We choose the following three recent pedestrian detection models trained on CityPerson (available in [6]): (i) Faster-RCNN [10] (ii) Cascade-RCNN [1] with the HRNet [11] backbone, and (iii) CSP-ResNet-50 [9].”
• Pedestrian Detection: The Elephant In The Room (2020, methodology): “5% MR − 2 better than CSP [23].”
• Improving Multispectral Pedestrian Detection by Addressing Modality Imbalance Problems (2020, methodology): “ed pedestrian detection has achieved notable progress in recent years with methods of occlusion handling [46,32,38], cascaded detection systems [30,2], semantic attention [49,3], anchor-free approach [31], etc. Nevertheless, current 4 Kailai Zhou, Linsen Chen, Xun Cao pedestrian detectors using sing”
• UA-DETRAC: A new benchmark and protocol for multi-object detection and tracking (2015, methodology): “…R-CNN (Girshick et al., 2014), CompACT (Cai et al., 2015), Faster R-CNN (Ren et al., 2017), EB (Wang et al., 2017), YOLOv3 (Redmon and Farhadi, 2018), GP-FRCNN (Amin and Galasso, 2017), CSP (Liu et al., 2019), HAT (Wu et al., 2019), FG-BR Net (Fu et al., 2019) and RD2 (Zhang et al., 2018) methods.”
• Temporal-Context Enhanced Detection of Heavily Occluded Pedestrians (2020, background): “With the renaissance of convolutional neural networks, many deep learning based methods on pedestrian detection [27, 28, 10, 26, 24, 23, 19, 17, 25, 18, 48, 36] significantly outperform the hand-crafted feature based methods [55, 61, 21, 14].”
• Soft Anchor-Point Object Detection (2019, methodology): “And it has been used in tasks such as scene text detection [34] and pedestrian detection [19].”
• BlockCopy: High-Resolution Video Processing with Block-Sparse Feature Propagation and Online Policies (2021, methodology): “Method Backbone Reasonable [test] Reasonable Bare Partial Heavy GMACs avg. sec/img CSP + BlockCopy (τ=0.3) ResNet-50 12.50 11.4 7.6 10.8 49.5 393 0.151 s MGAN + BlockCopy (τ=0.3) VGG-16 10.83 11.2 6.3 10.9 60.5 560 0.140 s CSP (CVPR2019) [25] ResNet-50 - 11.0 7.3 10.4 49.3 1128 † 0.330 s MGAN (ICCV2”