Getting human perception scores from street-level imagery.
The scores are in scale of 0-10.
Safety, lively, beautiful, wealthy high score indicates strong positive feeling
Boring, depressing high score indicates strong negative feeling
Model Accuracy：
| Model | safe | lively | wealthy | beautiful | boring | depressing |
|---|---|---|---|---|---|---|
| Accuracy | 76.7% | 77.1% | 72.9% | 76.9% | 61.6% | 67.2% |
The models are pre-trained on the MIT Place Pulse 2.0 dataset. The backbone of the models are vision transformer (ViT) pretrianed on ImageNet (ViT_B_16_Weights.IMAGENET1K_SWAG_E2E_V1).
In the ViT heads, 3 Linear layers with ReLU activiation are added for classification.
The models will be automatically downloaded when run eval.py (recommended method). You can also manually download the models here
Install packages from requirements.txt
pip install -r requirements.txt
Change the file path in eval.py
model_load_path = "./model" # path to save downloaded models
images_path = "./test_image" # input image path
out_Path = "./output" # output scores path
Run the file eval.py
python eval.py
Please cite our papers if you use this code or any of the models. Find more streetscapes here
@article{2024_global_streetscapes,
author = {Hou, Yujun and Quintana, Matias and Khomiakov, Maxim and Yap, Winston and Ouyang, Jiani and Ito, Koichi and Wang, Zeyu and Zhao, Tianhong and Biljecki, Filip},
doi = {10.1016/j.isprsjprs.2024.06.023},
journal = {ISPRS Journal of Photogrammetry and Remote Sensing},
pages = {216-238},
title = {Global Streetscapes -- A comprehensive dataset of 10 million street-level images across 688 cities for urban science and analytics},
volume = {215},
year = {2024}
}