Scale-Aware Fast R-CNN for Pedestrian Detection

By Xiaodan Liang, Shengmei Shen, Tingfa Xu, Jiashi Feng, Shuicheng Yan. 2015

837 citations recorded by the source corpus.

In this paper, we consider the problem of pedestrian detection in natural scenes. Intuitively, instances of pedestrians with different spatial scales may exhibit dramatically different features. Thus, large variance in instance scales, which results in undesirable large intracategory variance in features, may severely hurt the performance of modern object instance detection methods. We argue that this issue can be substantially alleviated by the divide-and-conquer philosophy. Taking pedestrian detection as an example, we illustrate how we can leverage this philosophy to develop a Scale-Aware Fast R-CNN (SAF R-CNN) framework. The model introduces multiple built-in subnetworks which detect pedestrians with scales from disjoint ranges. Outputs from all of the subnetworks are then adaptively combined to generate the final detection results that are shown to be robust to large variance in instance scales, via a gate function defined over the sizes of object proposals. Extensive evaluations on several challenging pedestrian detection datasets well demonstrate the effectiveness of the proposed SAF R-CNN. Particularly, our method achieves state-of-the-art performance on Caltech [P. Dollar, C. Wojek, B. Schiele, and P. Perona, “Pedestrian detection: An evaluation of the state of the art,” IEEE Trans. Pattern Anal. Mach. Intell., vol. 34, no. 4, pp. 743–761, Apr. 2012], and obtains competitive results on INRIA [N. Dalal and B. Triggs, “Histograms of oriented gradients for human detection,” in Proc. IEEE Conf. Comput. Vis. Pattern Recognit., 2005, pp. 886–893], ETH [A. Ess, B. Leibe, and L. V. Gool, “Depth and appearance for mobile scene analysis,” in Proc. Int. Conf. Comput. Vis., 2007, pp. 1–8], and KITTI [A. Geiger, P. Lenz, and R. Urtasun, “Are we ready for autonomous driving? The KITTI vision benchmark suite,” in Proc. IEEE Conf. Comput. Vis. Pattern Recognit., 2012, pp. 3354–3361].

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

Citations in the corpus

StairNet: Top-Down Semantic Aggregation for Accurate One Shot Detection (2017, background): “Many previous works argued that object scale difference causes different distribution in feature space [18, 39].”
An Analysis of Scale Invariance in Object Detection - SNIP (2017, background): “For example, features from the layers near to the input, referred to as shallow(er) layers, are combined with deeper layers for detecting small object instances [23, 34, 1, 13, 27], dilated/deformable convolution is used to increase receptive fields for detecting large objects [32, 7, 37, 8], indepe”
Bi-box Regression for Pedestrian Detection and Occlusion Estimation (2018, unknown): “Comparison with State-of-the-art Results In Figure 6, we compare our approach with some state-of-the-art approaches including DeepParts [31], CompACTDeep [6], SA-FastRCNN [19], MS-CNN [5], RPN+BF [39], F-DNN [10], FDNN+SS [10], PCN [34], JL-TopS [44], SDS-RCNN [4].”
Repulsion Loss: Detecting Pedestrians in a Crowd (2017, background): “Recently, the CNN-based detectors [17, 30, 21, 14, 28] show great potential in dominating the field of pedestrian detection.”
Predictive Inequity in Object Detection (2019, background): “Recent works have addressed this is by incorporating multiple representations into their model, one for each pedestrian scale (Li et al., 2018), or for different body parts to handle occlusion (Zhang et al., 2018).”
Unsupervised Hard Example Mining from Videos for Improved Object Detection (2018, background): “Many of these object detectors have been re-purposed for other tasks such as face detection [39,29,60,15], [31,63,62,25,57,23,66] and pedestrian detection [64,14,6,7,22,30,65], achieving impressive results [24,61,12].”
UnitBox: An Advanced Object Detection Network (2016, background): “This assumption violates the fact that the bounds of an object are highly correlated.”
Recent Advances in Deep Learning for Object Detection (2019, background): “Deep learning based methods for pedestrian detection [8, 9, 10, 205, 206, 207, 208, 209, 210, 211] showed excellent performance and achieved state-of-the-art results on public benchmarks.”
CityPersons: A Diverse Dataset for Pedestrian Detection (2017, background): “Recently good results are presented by customized pure convnet architectures such as MS-CNN [3] and SA-FastRCNN [21].”
In Teacher We Trust: Learning Compressed Models for Pedestrian Detection (2016, methodology): “We produce a model with only 157K parameters that outperforms AlexNet [19] which has over 57M parameters, 360 × as many.”
Occlusion-aware R-CNN: Detecting Pedestrians in a Crowd (2018, methodology): “[25] use multiple built-in sub-networks to adaptively detect pedestrians across scales.”
DC-SPP-YOLO: Dense Connection and Spatial Pyramid Pooling Based YOLO for Object Detection (2019, methodology): “[40] used the scale-aware mechanism to weight and combine the prediction results of Large-size Sub-network and Small-size Sub-network according to the size of input proposal, which got the state-of-the-art performance on pedestrian detection.”
Towards Large-Scale Small Object Detection: Survey and Benchmarks (2022, background): “Li et al. [74] built parallel subnetworks where small-size subnetwork is learned speciﬁcally to detect small pedestrians.”
Learning Efficient Single-Stage Pedestrian Detectors by Asymptotic Localization Fitting (2018, methodology): “SA-FastRCNN [24] jointly trains two networks to detect pedestrians of large scales and small scales respectively, based on the proposals generated from ACF detector [10].”
Multispectral Deep Neural Networks for Pedestrian Detection (2016, background): “This is encouraged by the observation in [21] that larger feature maps are beneﬁcial for detecting pedestrians of small image sizes.”
Refiner: Refining Self-attention for Vision Transformers (2021, background): “The locality (local receptive ﬁelds) and spatial invariance (weight sharing) have been proven to be the key of the success of CNNs across many computer vision tasks [31, 44, 53, 25, 1].”
Occluded Pedestrian Detection Through Guided Attention in CNNs (2018, unknown): “At the meantime, some works present good results using customized architectures, such as MS-CNN [4] and SA-FastRCNN [18].”