Laser & Optoelectronics Progress, Volume. 58, Issue 14, 1400002(2021)
Review of Computer Vision Based Object Counting Methods
Fig. 1. Schematic diagrams of three models. (a) Regression based object counting model; (b) density estimation based object counting model; (c) multi-task model
Fig. 2. Architecture of multi-scene judgment
Fig. 3. Architecture of FCN-rLSTM
Fig. 4. Input image and generation of density map. (a) Input image; (b) generation of density map
Fig. 5. Architecture of Hydra CNN
Fig. 6. Architecture of MCNN
Fig. 7. Architecture of DecideNet
Fig. 8. Structure of network of combined loss function
Fig. 9. Architecture of SaCNN
Fig. 10. Architecture of SFANet
Fig. 11. Architecture of CAT-CNN
Fig. 12. Architecture of FCN-MT
Fig. 13. Architecture of cell segmentation network
Fig. 14. Samples from six crowd datasets. (a) UCSD; (b) Mall; (c) UCF_CC_50; (d) WorldExpo’10; (e) Shanghai Tech Part A; (f) Shanghai Tech Part B
Fig. 15. Samples from three cell datasets. (a) VGG Cells; (b) MBM Cells; (c) Adipocyte Cells
Fig. 16. Samples from two datasets. (a) WebCamT; (b) TRANCOS
Fig. 17. Estimation results on Shanghai Tech dataset generated by SFANet. The first two rows belong to Part B, and the last two rows belong to Part A[58]. (a) Input images; (b) attention maps; (c) density maps; (d) ground truths
|
|
|
|
|
Get Citation
Copy Citation Text
Ni Jiang, Haiyang Zhou, Feihong Yu. Review of Computer Vision Based Object Counting Methods[J]. Laser & Optoelectronics Progress, 2021, 58(14): 1400002
Category: Reviews
Received: Oct. 10, 2020
Accepted: Dec. 3, 2020
Published Online: Jun. 30, 2021
The Author Email: Yu Feihong (feihong@zju.edu.com)