Bayesian Loss for Crowd Count Estimation With Point Supervision

By Zhiheng Ma, Xing Wei, Xiaopeng Hong, Yihong Gong. 2019

640 citations recorded by the source corpus.

In crowd counting datasets, each person is annotated by a point, which is usually the center of the head. And the task is to estimate the total count in a crowd scene. Most of the state-of-the-art methods are based on density map estimation, which convert the sparse point annotations into a “ground truth” density map through a Gaussian kernel, and then use it as the learning target to train a density map estimator. However, such a "ground-truth" density map is imperfect due to occlusions, perspective effects, variations in object shapes, etc. On the contrary, we propose Bayesian loss, a novel loss function which constructs a density contribution probability model from the point annotations. Instead of constraining the value at every pixel in the density map, the proposed training loss adopts a more reliable supervision on the count expectation at each annotated point. Without bells and whistles, the loss function makes substantial improvements over the baseline loss on all tested datasets. Moreover, our proposed loss function equipped with a standard backbone network, without using any external detectors or multi-scale architectures, plays favourably against the state of the arts. Our method outperforms previous best approaches by a large margin on the latest and largest UCF-QNRF dataset.

Subfields: Artificial Intelligence, Computer Vision and Pattern Recognition, Ocean Engineering

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

1. Xiaopeng Hong
2. Xing Wei
3. Xiaoyu Tao
4. Yihong Gong
5. Xinyuan Chang
6. Songlin Dong
7. Qi Wang
8. Xuelong Li

In Computer Vision and Pattern Recognition

1. Xiaopeng Hong
2. Xing Wei
3. Dacheng Tao
4. Xiaoyu Tao
5. Xinyuan Chang
6. Yihong Gong
7. Xuelong Li
8. Songlin Dong

In Ocean Engineering

1. Dimitris Samaras
2. Huidong Liu
3. Qi Wang
4. Junyu Gao
5. Wei Lin
6. Minh Hoai
7. Xuelong Li

Citations in the corpus

• Few-Shot Class-Incremental Learning (2020, background): “Convolutional Neural Networks (CNNs) have been successfully applied to a broad range of computer vision tasks [17, 11, 34, 4, 22, 25, 45, 19].”
• Distribution Matching for Crowd Counting (2020, methodology): “Most recent works [20, 35, 15, 4, 31, 50, 27, 40, 29, 26, 54, 8, 25, 55, 61, 59, 17, 48, 21, 23, 30, 47, 53, 43, 37, 56, 18, 39, 24, 44, 42] are based on density map estimation, which has been shown to be more robust than detection-then-count and count regression approaches.”
• Topology-Preserving Class-Incremental Learning (2020, methodology): “To date, deep neural networks have been successfully applied to a large number of computer vision and pattern recognition tasks [16, 11, 34, 33, 22, 5, 20, 8, 41, 40, 24], etc.”
• NWPU-Crowd: A Large-Scale Benchmark for Crowd Counting and Localization (2020, background): “Since BL [41] is supervised by point locations instead of density maps, PSNR and SSIM are not reported.”
• Attention Scaling for Crowd Counting (2020, methodology): “Inspired by the great success that the convolutional neural networks have made in computer vision tasks like object detection [28, 3, 27], image segmentation [4, 8], and object tracking [48, 47], it recently springs dozens of CNN based crowd counting methods [37, 22, 18, 41, 10].”
• Rethinking Counting and Localization in Crowds: A Purely Point-Based Framework (2021, background): “Although many compelling models have been proposed, these density map learning based models still fail to provide the exact locations of individuals in crowds, not to mention their inherent flaws as pointed out in [1, 27, 21].”
• Infrared-Visible Cross-Modal Person Re-Identification with an X Modality (2020, background): “2019) as many other computer vision tasks (Wang et al. 2010; Ma et al. 2019).”