Learning From Synthetic Data for Crowd Counting in the Wild

By Qi Wang, Junyu Gao, Wei Lin, Yuan Yuan. Computer Vision and Pattern Recognition, 2019

601 citations recorded by the source corpus.

Recently, counting the number of people for crowd scenes is a hot topic because of its widespread applications (e.g. video surveillance, public security). It is a difficult task in the wild: changeable environment, large-range number of people cause the current methods can not work well. In addition, due to the scarce data, many methods suffer from over-fitting to a different extent. To remedy the above two problems, firstly, we develop a data collector and labeler, which can generate the synthetic crowd scenes and simultaneously annotate them without any manpower. Based on it, we build a large-scale, diverse synthetic dataset. Secondly, we propose two schemes that exploit the synthetic data to boost the performance of crowd counting in the wild: 1) pretrain a crowd counter on the synthetic data, then finetune it using the real data, which significantly prompts the model's performance on real data; 2) propose a crowd counting method via domain adaptation, which can free humans from heavy data annotations. Extensive experiments show that the first method achieves the state-of-the-art performance on four real datasets, and the second outperforms our baselines. The dataset and source code are available at https://gjy3035.github.io/GCC-CL/.

Subfields: Artificial Intelligence, Computer Vision and Pattern Recognition

View Paper: https://www.semanticscholar.org/paper/3c667ab03a4e8842847f78227816d944a66afb32

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

Machine Learning for Synthetic Data Generation: a Review (2023, background): “Generally, synthetic data are defined as the artificially annotated information generated by computer algorithms or simulations [4], [12].”
JHU-CROWD++: Large-Scale Crowd Counting Dataset and A Benchmark Method (2020, methodology): “Wang et al.[44] introduced a synthetic crowd counting dataset that is based on GTA V electronic game.”
NWPU-Crowd: A Large-Scale Benchmark for Crowd Counting and Localization (2020, background): “GCC (synthetic) [20] 15,212 1080 1920 7,625,843 0 501 3,995 @ • @ •”
Synthetic Data for Deep Learning (2019, methodology): “[625] discuss synthetic-to-real domain adaptation in the context of crowd counting, a domain where synthetic data has been successfully used for a long time.”
Deep Structural Contour Detection (2020, neutral): “The head-density ground-truth of the datasets is provided by [43].”
Distribution Matching for Crowd Counting (2020, methodology): “Almost all prior crowd density map estimation methods [56, 57, 60, 38, 20, 33, 35, 4, 28, 50, 27, 40, 29, 26] have followed this convention.”
Attention Scaling for Crowd Counting (2020, methodology): “In this section, we evaluate our approach against sixteen currently reported methods [36, 5, 12, 19, 40, 32, 38, 31, 42, 20, 43, 6, 37, 22, 18, 41, 10] on ShanghaiTech Part A dataset [49], UCF CCF 50 dataset [10], UCF-QNRF Dataset [11] and WorldExpo’10 dataset [46].”
Generative artificial intelligence: a systematic review and applications (2024, neutral): “In another problem, Wang et al. [137] used the SSIM embedded-cycle GAN to count the number of people in the crowd.”