An object counting network based on hierarchical context and feature fusion

Object counting method based on dual attention network

“…As is shown in Table 3, the proposed method gets a 0.3% improvement of MAE with Li et al [19] and achieves the lowest 85.9 of MAE among them, except the method proposed in [15]. Compared to the method proposed in [15], the result of the proposed method on Shanghaitech_A is lower than it by 5.9% and the proposed method achieves higher performance on TRANCOS by 29.4%. Combining the comparison of TRANCOS and Shanghaitech_A, the proposed method has a competitive performance with the method proposed in [15].…”

“…We compare the proposed Count‐DANet with several counting methods and it achieves a significant improvement on different items in the GAME metric, and the detailed results are shown in Table 1. It can be deduced that the Count‐DANet has a 29.4% improvement of GAME(0) over Zhang et al [15] on this dataset and also outperforms other methods, except for the method proposed in [19]. Compared to the method proposed in [19], the proposed method achieves an inferior result than it and the value of the proposed method GAME(0) is only lower by 0.1 point, showing that the proposed method has a competitive performance with the method proposed in [19].…”

“…We compare our method with several typical crowd counting methods on Shanghaitech_A dataset: Marsden et al [29], Zhang et al [12], Sindagi and Patel [18], Sam et al [13], Li et al [16], SaCNN [14], Li et al [19], and Zhang et al [15], the results of which are shown in Table 3. As is shown in Table 3, the proposed method gets a 0.3% improvement of MAE with Li et al [19] and achieves the lowest 85.9 of MAE among them, except the method proposed in [15]. Compared to the method proposed in [15], the result of the proposed method on Shanghaitech_A is lower than it by 5.9% and the proposed method achieves higher performance on TRANCOS by 29.4%.…”

“…Zhang et al [14] proposed an object counting method based on feature fusion to address the scale variant problem. Similarly, Zhang et al [15] also developed a cascaded network based on feature fusion for object counting. Based on the cascaded architecture in [15], Li et al [16] designed a multi‐scale structure similar to U‐Net to cope with the scale‐variant issue.…”

“…Similarly, Zhang et al [15] also developed a cascaded network based on feature fusion for object counting. Based on the cascaded architecture in [15], Li et al [16] designed a multi‐scale structure similar to U‐Net to cope with the scale‐variant issue. Simultaneously, though some researchers attempted to add the context information from the images to the CNN architecture to boosting the counting accuracy, these methods cannot solve the multi‐scale problem.…”