This folder contains of a demo test code that can be used for DM Count network. The original testing code was taken from the DM Count repo and modified.
The pretrained models that were used to prepare results for different datasets and networks are given in the following zenodo links.
| Network name | link |
|---|---|
| BL (NWPU,UCF,STA,STB) | here |
| DM Count (NWPU,UCF,STA,STB) | here |
| SCARNet (NWPU,UCF,STA,STB) | here |
| SDCNet (NWPU,UCF,STA,STB) | here |
| SFANet (NWPU,UCF,STA,STB) | here |
Code to test the model :
python test.py --dataset <nwpu,qnrf,sta,stb> --model-path <local path to the stored model> --data-path <local path of the data location>
This code generates a txt file with the name img_name_targ_pred.txt this file contains the list of counts as :
image_name_1 , target_1, predicted_1
image_name_2 , target_2, predicted_2
.
.
.
image_name_n , target_n, predicted_n
Here image_name_i can be a string or an int that refers to the input image name to the network and predicted_i refers to the count predicted by your network from the image_name_i, and target_i represents its ground truth.
We encourage you to send us a pull request here following the format mentioned here to include your network in our work.