Deep learning

Mathematics

Bounding overwatch

Regression

Image (mathematics)

Statistics

Machine learning

Minimum bounding box

Ground truth

Artificial intelligence

Geometry

Crowds

Computer science

Point (geometry)

Computer security

Artificial neural network

Modern crowd counting methods usually employ deep neural networks (DNN) to estimate crowd counts via density regression. Despite their significant improvements, the regression-based methods are incapable of providing the detection of individuals in crowds. The detection-based methods, on the other hand, have not been largely explored in recent trends of crowd counting due to the needs for expensive bounding box annotations. In this work, we instead propose a new deep detection network with only point supervision required. It can simultaneously detect the size and location of human heads and count them in crowds. We first mine useful person size information from point-level annotations and initialize the pseudo ground truth bounding boxes. An online updating scheme is introduced to refine the pseudo ground truth during training; while a locally-constrained regression loss is designed to provide additional constraints on the size of the predicted boxes in a local neighborhood. In the end, we propose a curriculum learning strategy to train the network from images of relatively accurate and easy pseudo ground truth first. Extensive experiments are conducted in both detection and counting tasks on several standard benchmarks, e.g. ShanghaiTech, UCF_CC_50, WiderFace, and TRANCOS datasets, and the results show the superiority of our method over the state-of-the-art.

Point in, Box Out: Beyond Counting Persons in Crowds