Offset-decoupled deformable convolution for efficient crowd counting

Abstract

Crowd counting is considered a challenging issue in computer vision. One of the most critical challenges in crowd counting is considering the impact of scale variations. Compared with other methods, better performance is achieved with CNN-based methods. However, given the limit of fixed geometric structures, the head-scale features are not completely obtained. Deformable convolution with additional offsets is widely used in the fields of image classification and pattern recognition, as it can successfully exploit the potential of spatial information. However, owing to the randomly generated parameters of offsets in network initialization, the sampling points of the deformable convolution are disorderly stacked, weakening the effectiveness of feature extraction. To handle the invalid learning of offsets and the inefficient utilization of deformable convolution, an offset-decoupled deformable convolution (ODConv) is proposed in this paper. It can completely obtain information within the effective region of sampling points, leading to better performance. In extensive experiments, average MAE of 62.3, 8.3, 91.9, and 159.3 are achieved using our method on the ShanghaiTech A, ShanghaiTech B, UCF-QNRF, and UCF_CC_50 datasets, respectively, outperforming the state-of-the-art methods and validating the effectiveness of the proposed ODConv.

A Dynamic Network with Transformer for Image Denoising

“…In line with the input, the final parameters used for the convolution kernel vary [27]. However, these studies only focus on the information of a single one-dimensional convolution kernel, ignoring information including convolution size, and number of input and output channels [29]. To adopt all of the dimensional information, omni-dimensional dynamic convolution (OD-Conv) utilizes a multidimensional attention mechanism to learn convolutional kernels from four dimensions and applies these attention weights to the corresponding convolutional kernels [30].…”

Apple Grading Based on Multi-Dimensional View Processing and Deep Learning

“…After adding the Retinex algorithm, the influence of illumination on apple surface features could be effectively reduced, and the model detection accuracy was improved by 4.65%. After adding ODConv convolution, the network backbone could enhance the extraction ability of apple surface features, and the detection accuracy of the model was improved by 2.64%, but the number of model parameters was increased by 1.92 M [21]. After the overall lightweight improvement of the neck, the model detection accuracy was 96.56%; only 0.42% of the detection accuracy was sacrificed, but the number of model parameters was reduced by 2.54 M, which optimized the model structure.…”