This is a Python script to find height, weight, segmentation mask and joint locations of a person from a full-body single person image. Used networks are implemented in "Height and Weight Estimation From Unconstrained Images" paper. You can find this paper's repository here.
If you find this program useful in your research, please consider citing:
@inproceedings{altinigne2020height,
title={Height and Weight Estimation from Unconstrained Images},
author={Altinigne, Can Yilmaz and Thanou, Dorina and Achanta, Radhakrishna},
booktitle={ICASSP 2020-2020 IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP)},
pages={2298--2302},
year={2020},
organization={IEEE}
}
Each file is nearly 2GB. Just download them and put them in /models folder.
- Pretrained Height Network: https://drive.google.com/open?id=1fX0DDgbTcOOmiz9KdtU7I2YYg5S49upj
- Pretrained Weight Network: https://drive.google.com/open?id=14ShT0rsUohiGT0wJlKY9cGHgEy0w4Ity
- PyTorch = 1.2.0
- Python = 3.6
In order to get outputs, just run python HWFinder.py -i [IMAGE ADDRESS] -g [GPU NUMBER] -r [RESOLUTION]. The models are trained using images with a resolution of 128x128, so set the RESOLUTION parameter between 128 and 256.
You can use environment.yml file to create a Conda environment to run the script. You can create a new environment using this command. conda env create -f environment.yml -p [PATH]