Detects the visual condition of a solar panel from the uploaded images in runtime
Classes of Data
- Clean
- Dusty
- Bird Drop
- Electrical Damage
- Physical Damage
- Snow Covered
Suggests remedies or suggestive measures for each type of anomaly
Accuracy : 100%
Dataset Link: https://www.kaggle.com/datasets/pythonafroz/solar-panel-images
NOTE: Can also detect images from Google. But in case of outside images, the sensitivity increases for "Clear" class
- download the files (click on Green "Code" button on top right of the files area in Github)
- extract the files
- open with VScode or Spyder or any IDE of your preference (Not Jupyter Notebooks)
- install the packages that are listed in requirements.txt
- open a terminal in your IDE or in CMD and type
streamlit run app.py - type your email in the terminal if you are using streamlit for the first time
- use the app
- to host, use streamlit services (As Heroku services are discotinued in Nov, 22)