varun-projects / rooftop-solar-potential-detection-using--satelite-imagery

We are going to map the location and where we can install Solar Panels on rooftops. To achieve our required results, we are going to use supervised methods based on CNN

ROOFTOP-SOLAR-POTENTIAL-DETECTION-USING--SATELITE-IMAGERY

As we know nonrenewable energy is not a good thing to be dependent on it because of its harmful effect on the environment caused by the greenhouse gases and now we can also see the rising cost of the petrol a fossil fuel. So, what we can do about it? we can try to shift our attention towards the renewable source of energy. This is where we come into picture we are finding the potential surface area on rooftops to install photovoltaics (PV) panels which can be really useful because of solar energy and calculating the solar energy generated by a particular area.

We are achieving this by applying Machine Learning techniques with satellite and aerial imagery, will help in bypassing the limitations of surveys in providing the surveys in providing this mapping at large scale. Being more specific we are using convolutional neural network to describe available rooftop area on which we can install photovoltaics (PV) panels by the method of pixel wise segmentation. We picked up the dataset form the Kragel which was labelled. We tried different data augmentation by flipping data 90 degrees, rotating it in order to increase the model performance. After training our model on the dataset we are able to detect the rooftop with the accuracy on the test set of about 90.33% and an Intersection over Union (IoU) of 55.65% with only 245 images in the training dataset. In future our model will able to predict the available rooftop area with more accuracy and on different geographical locations as we know different geographical location have different style of rooftops so we have to train our model on that type of data then it will able to predict the available rooftop area.