在如今社會，公共場景的安全防範任務得到了極大重視。眾多技術的出現，也使得我們的管控更為有效。人群計數是其中一項重要的技術，其可以幫助我們估算當前場景中的人群密集程度從而防範危險事故的發生。
本文使用了基於全卷積神經網絡的模型來進行人群計數，並且探索了結果圖片大小對模型性能的影響。我們嘗試了切片的方法來增加訓練時每批次的圖片數量，並且改進了基於標註的熱度圖生成方式，最後在ShanghaiTech part A和UCF-QNRF等資料集上達到了領先的效果。

In modern society, the security problem in public scenes has gained great interests. With the rapid advances in computer vision and deep learning, we have achieved significant progress in video surveillance technology. Crowd counting is one of the important problems, which is to estimate the total number of people in a specific scene from an image for better security control.
This thesis uses a simple and efficient network based on fully convolutional network and exploits the effect of the size of the output. We enlarge the batch size by cropping the image and improve the heatmap generation method. Finally, we archive state-of-the-art result in Shanghaitech part A and UCF-QNRF.

1 Introduction 1
1.1 Motivation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1
1.2 Problem Description . . . . . . . . . . . . . . . . . . . . . . . . . 1
1.3 Contributions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3
1.4 Thesis Organization . . . . . . . . . . . . . . . . . . . . . . . . . . 4
2 Related Work 5
2.1 Detection based approach . . . . . . . . . . . . . . . . . . . . . . . 5
2.2 Density estimation based approach . . . . . . . . . . . . . . . . . . 6
2.2.1 Heatmap Generation . . . . . . . . . . . . . . . . . . . . . 6
2.2.2 Regression Approach . . . . . . . . . . . . . . . . . . . . . 7
2.3 Datasets for Crowd Counting . . . . . . . . . . . . . . . . . . . . . 9
2.3.1 Shanghaitech Dataset . . . . . . . . . . . . . . . . . . . . . 9
2.3.2 UCF-QNRF Dataset . . . . . . . . . . . . . . . . . . . . . 10
3 Proposed Method 12
3.1 Hierarchical geometry-adaptive kernel . . . . . . . . . . . . . . . . 12
3.2 Network . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 13
3.3 Scale Factor . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 15
3.4 Large Batch Size . . . . . . . . . . . . . . . . . . . . . . . . . . . 16
4 Experimental Results 18
4.1 Training Setting . . . . . . . . . . . . . . . . . . . . . . . . . . . . 18
4.2 Data Augmentation . . . . . . . . . . . . . . . . . . . . . . . . . . 18
4.3 Loss Function . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 18
4.4 Evaluation Method in Crowd Counting Task . . . . . . . . . . . . . 19
4.4.1 Dataset Splitting . . . . . . . . . . . . . . . . . . . . . . . 19
4.4.2 Evaluation Metrics . . . . . . . . . . . . . . . . . . . . . . 19
4.5 Experiment on Shanghaitech A . . . . . . . . . . . . . . . . . . . . 20
4.6 Ablation Study on Shanghaitech A . . . . . . . . . . . . . . . . . . 21
4.7 Experiment on UCF-QNRF . . . . . . . . . . . . . . . . . . . . . . 23
4.8 Case Analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 25
4.8.1 Cases for Relative Error . . . . . . . . . . . . . . . . . . . 25
4.8.2 Cases for Absolute Error . . . . . . . . . . . . . . . . . . . 27
5 Conclusions 29
References 30

[1] J. Long, E. Shelhamer, and T. Darrell, “Fully convolutional networks for semantic segmentation,” in Proceedings of the IEEE conference on computer
vision and pattern recognition, pp. 3431–3440, 2015.
[2] Y. Zhang, D. Zhou, S. Chen, S. Gao, and Y. Ma, “Single-image crowd counting
via multi-column convolutional neural network,” in Proceedings of the IEEE
conference on computer vision and pattern recognition, pp. 589–597, 2016.
[3] H. Idrees, M. Tayyab, K. Athrey, D. Zhang, S. Al-Maadeed, N. Rajpoot, and
M. Shah, “Composition loss for counting, density map estimation and localization in dense crowds,” in Proceedings of the European Conference on Computer Vision (ECCV), pp. 532–546, 2018.
[4] N. Liu, Y. Long, C. Zou, Q. Niu, L. Pan, and H. Wu, “Adcrowdnet: An
attention-injective deformable convolutional network for crowd understanding,” arXiv preprint arXiv:1811.11968, 2018.
[5] C. Zhang, H. Li, X. Wang, and X. Yang, “Cross-scene crowd counting via
deep convolutional neural networks,” in Proceedings of the IEEE conference
on computer vision and pattern recognition, pp. 833–841, 2015.
[6] X. Cao, Z. Wang, Y. Zhao, and F. Su, “Scale aggregation network for accurate
and efficient crowd counting,” in Proceedings of the European Conference on
Computer Vision (ECCV), pp. 734–750, 2018.
[7] H. Idrees, I. Saleemi, C. Seibert, and M. Shah, “Multi-source multi-scale
counting in extremely dense crowd images,” in Proceedings of the IEEE conference on computer vision and pattern recognition, pp. 2547–2554, 2013.
[8] Y. Li, X. Zhang, and D. Chen, “Csrnet: Dilated convolutional neural networks
for understanding the highly congested scenes,” in Proceedings of the IEEE
conference on computer vision and pattern recognition, pp. 1091–1100, 2018.
[9] P. Dollar, C. Wojek, B. Schiele, and P. Perona, “Pedestrian detection: An evaluation of the state of the art,” IEEE transactions on pattern analysis and machine intelligence, vol. 34, no. 4, pp. 743–761, 2011.
[10] S. Yang, P. Luo, C. C. Loy, and X. Tang, “Wider face: A face detection
benchmark,” in IEEE Conference on Computer Vision and Pattern Recognition (CVPR), 2016.
[11] S. Ren, K. He, R. Girshick, and J. Sun, “Faster r-cnn: Towards real-time object
detection with region proposal networks,” in Advances in neural information
processing systems, pp. 91–99, 2015.
[12] W. Liu, D. Anguelov, D. Erhan, C. Szegedy, S. Reed, C.-Y. Fu, and A. C. Berg,
“Ssd: Single shot multibox detector,” in European conference on computer
vision, pp. 21–37, Springer, 2016.
[13] P. Hu and D. Ramanan, “Finding tiny faces,” in Proceedings of the IEEE conference on computer vision and pattern recognition, pp. 951–959, 2017.
30
[14] T. Song, L. Sun, D. Xie, H. Sun, and S. Pu, “Small-scale pedestrian detection based on somatic topology localization and temporal feature aggregation,”
arXiv preprint arXiv:1807.01438, 2018.
[15] V. Lempitsky and A. Zisserman, “Learning to count objects in images,” in
Advances in neural information processing systems, pp. 1324–1332, 2010.
[16] L. Boominathan, S. S. Kruthiventi, and R. V. Babu, “Crowdnet: A deep convolutional network for dense crowd counting,” in Proceedings of the 24th ACM
international conference on Multimedia, pp. 640–644, ACM, 2016.
[17] I. Goodfellow, J. Pouget-Abadie, M. Mirza, B. Xu, D. Warde-Farley, S. Ozair,
A. Courville, and Y. Bengio, “Generative adversarial nets,” in Advances in
neural information processing systems, pp. 2672–2680, 2014.
[18] S. Jiang, X. Lu, Y. Lei, and L. Liu, “Mask-aware networks for crowd counting,”
arXiv preprint arXiv:1901.00039, 2018.
[19] X. Wu, Y. Zheng, H. Ye, W. Hu, J. Yang, and L. He, “Adaptive scenario discovery for crowd counting,” in ICASSP 2019-2019 IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP), pp. 2382–2386,
IEEE, 2019.
[20] W. Liu, M. Salzmann, and P. Fua, “Context-aware crowd counting,” in Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition,
pp. 5099–5108, 2019.
[21] K. Simonyan and A. Zisserman, “Very deep convolutional networks for largescale image recognition,” arXiv preprint arXiv:1409.1556, 2014.
[22] S. Ioffe and C. Szegedy, “Batch normalization: Accelerating deep
network training by reducing internal covariate shift,” arXiv preprint
arXiv:1502.03167, 2015.
[23] H. Noh, S. Hong, and B. Han, “Learning deconvolution network for semantic
segmentation,” in Proceedings of the IEEE international conference on computer vision, pp. 1520–1528, 2015.
[24] S. Ruder, “An overview of gradient descent optimization algorithms,” arXiv
preprint arXiv:1609.04747, 2016.
[25] D. P. Kingma and J. Ba, “Adam: A method for stochastic optimization,” in
3rd International Conference on Learning Representations, ICLR 2015, San
Diego, CA, USA, May 7-9, 2015, Conference Track Proceedings, 2015.
[26] C. Wang, Y. Wang, Z. Lin, A. L. Yuille, and W. Gao, “Robust estimation of 3d
human poses from a single image,” in Proceedings of the IEEE Conference on
Computer Vision and Pattern Recognition, pp. 2361–2368, 2014.
[27] M. Everingham, L. Van Gool, C. K. I. Williams, J. Winn, and A. Zisserman,
“The pascal visual object classes (voc) challenge,” International Journal of
Computer Vision, vol. 88, pp. 303–338, June 2010.
31
[28] T.-Y. Lin, M. Maire, S. Belongie, J. Hays, P. Perona, D. Ramanan, P. Dollár,
and C. L. Zitnick, “Microsoft coco: Common objects in context,” in European
conference on computer vision, pp. 740–755, Springer, 2014.
[29] J. Deng, W. Dong, R. Socher, L.-J. Li, K. Li, and L. Fei-Fei, “ImageNet: A
Large-Scale Hierarchical Image Database,” in CVPR09, 2009.
[30] Q. Wang, J. Gao, W. Lin, and Y. Yuan, “Learning from synthetic data for crowd
counting in the wild,” in Proceedings of the IEEE Conference on Computer
Vision and Pattern Recognition, pp. 8198–8207, 2019.
[31] X. Jiang, Z. Xiao, B. Zhang, X. Zhen, X. Cao, D. Doermann, and L. Shao,
“Crowd counting and density estimation by trellis encoder-decoder networks,”
in Proceedings of the IEEE Conference on Computer Vision and Pattern
Recognition, pp. 6133–6142, 2019.