The problem Shiwu solves

The problem Shiwu solves
Technology is supposed to be utilised for the greater good and our project aims to cover three vital sustainable development goals. When technology started to exist, it did because of one commmon goal - Technology for the greater good. Since childhood, we are being taught the principles of sustainable development and ways we can revolutionise the world and take care of our planet.

Zero Hunger: With the knowledge of the right crop for your soil, more crops can be produced thereby ending hunger.
Responsible Consumption and Production: We detect the correct soil and suggest the best crops which can be grown thereby ensuring sustainable production patterns.
Life on Land: When the right crop is being grown on the soil, the soil resources are utilised properly thereby preventing desertification and trying to halt desertification.
How does this work?
We used a soil classification dataset with four different types of soil - alluvial, black, clay and red soil which was available on Kaggle. It contains both train and test datasets for the different soil types. We have 5 hidden layers in the model. Our first layer is sequential as there's only a single input. Since it is a picture we need to flatten the RGB content. We have then used batch normalisation layer to normalise the mean output and standard deviation. Then we have densely connected 4 layers. Again, to normalise the output we have used batch normalisation and we have used the activation function 'softmax'. We have compiled the whole model using optimiser 'Adam' , handled losses using 'categoricalCrossentropy' and using metrics 'accuracy'. We have set up 50 epochs for better accuracy. Then we used flask to integrate it with our website,

Who can use this tech?
Farmers and even people who wish to have an in-house garden at their place can utilise this software and get the best suggestions for their next produce.

Challenges I ran into

Identification of the correct model and training and testing it with a machine of the bare minimum power took us a little longer than usual.
We also faced issues with deploying our model due to lack of disk space but we overcame those problems together as a team.