The problem Project garbAIge solves

In today’s world, garbage is a thing that can be seen everywhere in our surroundings and we are producing waste all the time 24/7hours & and it is continuously growing over time.

According to stats, India produces around 62 million tonnes of waste per year, which is a considerable number & to decrease the number of wastage, we have to follow and understand the importance of the waste segregation process to recycle maximum waste into valuable products. 

Waste segregation is a process under which the garbage or waste is separated into different groups. These waste groups are created by looking at the characteristics of the waste; similar categories of waste are collected in the same groups. 

Can we use AI to solve the problem of Waste Classification?

Project garbAIge aims to address that issue by reducing human contact to ever growing waste both unhygienic and toxic in nature and aims towards helping the planet take its first step towards sustainable living.

Project garbAIge automates to process of garbage segregation and promote recycling of appropriate classes of trash produced by humans.

Challenges I ran into

The first challenge was to ather all the libraries required for the completion of the project and deciding the frontend framework to use.

Gathering the dataset required extensive research on Kaggle

In order to choose the appropriate method for training the model, it was required to research on existing literature and see if any improvements or implementations could be done in a short span of time

Since the trained model was large in size, Github did not allow the model to be pushed by default, in this process, I learnt about Git LFS and configured the same on my system and went ahead with pushing the complete code.

Deploying on frontend was also challenging since it was required to have a "requirements.txt" and proper branch of the Github repository was to be selecte and taken care of.

Future Work includes:

Add more garbage classes and support multiclass classification

Enable real-time trash classification

Integrate the technology with IoT Devices and enable the same to be industry ready for autonomous garbage collection and segregation

Make the model more robust by augmenting the dataset and increasing the size of the dataset.