PE-PM2.5

Image-based PM2.5 Estimation From Imbalanced Data Distribution Using Prior-Enhanced Neural Networks

Dependencies

Python 3.8 PyTorch 1.12.1 opencv-python 4.7.0.72

Run

Put your test images in the folder 'imgs/', and then run the following scripts.

# change parameters in the test.py as you like
python test.py

Checkpoints & Datasets

Google Or Baidu

Useful links

Image-based-PM2.5-Estimation

Citing

The code is free for academic/research purpose. Please kindly cite our work in your publications if it helps your research.

@article{fang2024image,
  title={Image-Based PM2.5 Estimation From Imbalanced Data Distribution Using Prior-Enhanced Neural Networks},
  author={Fang, Xueqing and Li, Zhan and Yuan, Bin and Chen, Yihang},
  journal={IEEE Sensors Journal},
  volume={24},
  number={4},
  pages={4677--4693},
  year={2024},
  publisher={IEEE},
  link={10.1109/JSEN.2023.3343080}
}

Example

Model	img_1 (left)	img_2 (right)
Ground Truth	20.00	73.00
PE-ResNet18	23.16	64.03

Model	img_1 (left)	img_2 (right)
Ground Truth	80.0	152.0
PE-ResNet18	84.84	125.80

Model	img_1 (left)	img_2 (right)
Ground Truth	22.0	186.0
PE-ResNet18	25.85	184.47

lizhangray/PE-PM2.5