SCAR: Spatial-/channel-wise attention regression networks for crowd counting

In recent years, deep learning based crowd counting networks have achieved significant progress. However, most of them generate rough crowd density maps due to low-resolution features used for estimating crowd distribution, which affects the performance of crowd counting. To solve this problem, in this paper, we propose a Hierarchical Attention Guided Network (HAGN) for crowd counting. We apply the first 13 layers of VGG-16 to extract base features. Then, the extracted features are processed by the Hierarchical Attention Mechanism (HAM), which guided the extracted features to enlarge step by step via our proposed attention guided branch. Finally, the outputs of HAM are fed to 1 × 1 convolutional layer for final crowd density estimation. Experiments are performed on ShanghaiTech and UCF-QNRF datasets, and our HAGN achieves promising performance compared with the other state-of-the-art methods on crowd counting and crowd localization, respectively.

show abstract

“…Attention modules are often used in visual tasks [60]. Mnih et al [33] adds attention information to the RNN-based network firstly.…”

Section: B Attention Modulementioning

confidence: 99%

HAGN: Hierarchical Attention Guided Network for Crowd Counting

2020

View full text Add to dashboard Cite

show abstract

“…SANet [27] used the convolution kernel of various scales to deal with crowd scale variation. With the extensive application of attention mechanism based deep learning models in various computer vision tasks such as CBAM [14], SEblock [18] and non-local attention mechanism [16] had been successfully applied in crowd counting tasks [4] [50]. In addition, self-supervised and unsupervised methods also had important applications in crowd counting tasks [5] [19].…”

Section: A Crowd Countingmentioning

confidence: 99%

AAFM: Adaptive Attention Fusion Mechanism for Crowd Counting

2020

View full text Add to dashboard Cite

CNN-based crowd counting methods have achieved great progress in recent years. However, most of these CNN-based crowd counting methods do not make full use of contextual information, which contains high-level semantic features and low-level detail features from different receptive fields of CNN. But rich contextual information is important to solve the scale variation problem of crowd counting. So the precision of previous CNN-based crowd counting methods is decreased. To solve this problem, we propose an adaptive attention fusion mechanism (AAFM). AAFM can use multi-scale features from different receptive fields of CNN effectively. It integrates the convolution network for feature learning and the attention mechanism for multi-scale features fusion. We apply the first 13 convolution layers of VGG-16 as the encoder module to extract the base features. Then, the base features are fed into the decoder module. The decoder module mainly contains Density Regression Branch (DRB) and Feature Fusion Branch (FFB). DRB uses multiple convolution layers for feature learning and multi-scale feature extraction. FFB uses attention modules for modeling multi-scale features and element-wise multiply for features fusion. Therefore, AAFM can obtain rich contextual information into the encoder-decoder framework for generating high-quality crowd density maps and accurate counting. We perform experiments on ShanghaiTech, UCF-CC-50, and UCF-QNRF datasets, and AAFM achieves promising results.

show abstract

“…Specifically, S-DCNet defines and interval partition of [0, +∞) as {0}, (0, C1], (C2, C3], ..., (CM−1, CM] and (CM, +∞). These M+1 sub-intervals are labelled to the 0-th to N-th classes, respectively (Gao et al, 2019). The value of CM is set such that it is not greater than max local count of the training dataset.…”

Section: Spatial Divide and Conquer Networkmentioning

confidence: 99%

“…In this work, we mainly compare the analogy and divergence of different networks used in Convolutional Neural Network i.e. SCAR (Spatial/Channel-wise Attention Regression Network) (Gao et al, 2019), SDC-Net (Spatial Divide and Conquer) and CSR-Net (Xiong et al, 2020;.…”

Section: Introductionmentioning

confidence: 99%