The problem Waste Segregation solves

We are working on the poblem of Waste Management. We have observed that a lot cities, states and even countries faces problem on how to manage waste. Waste Management system have a golden 3R rule Recycle Reuse Reduce. But, it becomes more effecient if we include also consider the classification in Biodegradable and Non Biodegradable. As the Biodegradable waste can be used to produce biogas and various useful side products.

Here, we are making Deep Learning Model of Waste segregation which will identify the type of waste i.e., Biodegradable, Non-Biodegradable (Sub class: Recyclable, Resuable). This model then can be deployed at the dumping area, where the waste will be rolled on the conveyer-belt, then the Mechanic machine can segregate the waste. Since, the mechanic model is not possible in online mode, we will be deploying this model on a website, where the images of waste can be uploaded, and the model classify it.

Challenges we ran into

Challenges we faced into:

1. Facing trouble for setting hyperparameters for tuning model
2. Errors while deploying on heroku as the model weights were very large, we then modified our model architecture and reduced the size of model.
3. Very tedious error occurred whille integrating model app with the frontend app. First was of CORS which needed to be used while using React with Flask and other was transfering the data between two apps.