CommissarMa / Context-Aware_Crowd_Counting-pytorch

The implementation of Context-Aware Crowd Counting(CVPR2019)
MIT License
68 stars 23 forks source link

Just resize the img and target to the desired size? #22

Open Harr7y opened 4 years ago

Harr7y commented 4 years ago

resize the img and target to the desired size? It's not a common approach and not a ideal handle intuitively. The image size of SHT-A is quite Variegated, will that make the MAE bigger?