The problem SpyTrash solves

As the population in a society grows, there is a significant need to efficiently manage the waste produced and dispose of it. On a large scale, waste segregation isn’t used efficiently, instead being dumped into landfills and dumping grounds. In India, 77% of waste is dumped in landfills without any treatment, segregation, or safety checks. Not only is this wasteful, but it takes up a lot of land that could be used to grow and sustain India’s growing economy. Not segregating/separating waste leads to a tremendous amount of waste, unhygienic conditions, and spreads disease and illness. This provides a very unique opportunity to not only provide a very efficient waste management system structured around CV technology, while also informing the general public about the crisis, along with how to best handle waste in their personal lives, community, industry, and municipality.

Challenges we ran into

We had several issues with generating and loading a model in the correct format into the front end python file. As the model was in the format of a h5 model, the integration of that into the front end Flask app required the correct imports along with the compatible versions of Tensorflow and Keras. This solution required several hours to understand and research, as the versions of tensorflow on the model and the application were different, and some methodologies were not available to be synced. So, using the appropriate virtual environments and installations of the packages in the front end application, we were able to successfully load the instance of the model into the front end application and use the required functions, labels, and functionalities. We solved this problem by researching the documentation of the Keras framework, along with the documentation on the various methods used in the framework, and the integration of the Flask app to the backend.