This algorithm estimate the crowd density pixel-by-pixel using the vgg model.
- Input: images downloaded from the minio server's ucs-alg bucket
- Output: a crowd density matrix of the same size as the input image, saved as file and uploaded to the minio server
By default, the model is fixed, thus the model selection will not take effects.
- ucs_alg_node-0.1.6 or above
- numpy
- torch
- torchvision
- pillow
see main.py