INTRA-HOUR SOLAR IRRADIANCE ESTIMATION USING INFRARED SKY IMAGES AND MOBILENETV2-BASED CNN REGRESSION

Code Organization

All codes are written in python.

• main.ipynb Trains the model. Consists of Data Loaders, Model, Training and Evaluation cells. Uses the scrips stored in the .scripts/ folder

• prepareDs.ipynb Prepares the dataset in a trainable format.

.scripts/

• saveResults.py Calculates loss, Creates results folder for particular training based on hyperparameters and time, Saves the results and diagrams.

• imagetoTensor.py converts image to pytorch tensor for testing purposes.

• saveThermalImage.py converts grayscale infrared image to thermal image using colormap. Used for the menuscript.

.results/

The output best_model, results and diagrams are stored here.