Multi‐level feature fusion network for crowd counting

Abstract

Crowd counting has become a noteworthy vision task due to the needs of numerous practical applications, but it remains challenging. State‐of‐the‐art methods generally estimate the density map of the crowd image with the high‐level semantic features of various deep convolutional networks. However, the absence of low‐level spatial information may result in counting errors in the local details of the density map. To this end, a novel framework named Multi‐level Feature Fusion Network (MFFN) for single image crowd counting is proposed. The proposed MFFN, which is constructed in an encoder–decoder fashion, incorporates semantic and spatial information for generating high‐resolution density maps of input crowd images. Skip connections are developed between the encoder and the decoder so that low‐level spatial information and high‐level semantic features can be combined by element‐wise addition. In addition, a dense dilated convolution block is placed behind the encoder, extracting multi‐scale context features to guide feature fusion by a channel attention mechanism. The model is trained by multi‐task learning; semantic segmentation supervision is introduced to enhance feature representation. Extensive experiments are conducted on three crowd counting datasets (ShanghaiTech, UCF_CC_50, UCF‐QNRF), and the results show that MFFN outperforms state‐of‐the‐art methods. In addition, sufficient ablation studies are performed to verify the effectiveness of each component in our proposed method.

Art Image Style Conversion Based on Multi-Scale Feature Fusion Network

“…It was superior to existing methods. The ablation study confirmed the effectiveness of each component [7]. Shen et al built a hyper-spectral classification strategy ground on a three-dimensional MSFF strategy and channel attention mechanism to address the difficulties of traditional 2D or 3D deep CNNs in hyper-spectral image classification.…”

EMDL-ac4C: identifying N4-acetylcytidine based on ensemble two-branch residual connection DenseNet and attention

“…The second was where to place the attention mechanism in the model. There are various options for where to insert attention into the model, including introducing it in densecells ( Bastings and Filippova, 2020 ), adding it in denseblocks ( Wei et al, 2019 ; Zhou et al, 2019 ; Wang et al, 2021 ), inserting it between denseblocks and transitions ( Jia et al, 2022 ; Jia et al, 2023 ), bring in the attention mechanism in the transition layer ( Song et al, 2021 ), or attaching it before the data enters DenseNet or at the end of the model prediction.…”

An innovative development model for combining visual communication and art design based on the integration of multilevel features

“…Based on the coarse extraction of features in Section 2. H can be obtained by fusing three depth-level features from the VGG16 pillar network [16].…”