Deep neural network deep learning
Budget: ₹1,500 – ₹12,500 INR
The purpose is to assist people in better recycling efforts and to reduce recycling contamination. So, while you may not be explicitly directed to develop a full-fledged software program, you are expected to develop a deep learning model that could serve as a critical component in such a software application.
Here's a breakdown based on the tasks:
Task 1: Smart Recycling using Deep Learning
Task 1.1: Define an image classification problem that could help with recycling efforts. You will have to identify the classes of recyclable items, like paper, plastic, metal, etc., and non-recyclable items.
Task 1.2: Make a plan for data collection, how you'll split the data, and how you'll evaluate the model's performance.
Task 1.3: Implement the deep learning model using the dataset you've gathered or identified.
Task 2: Analyse and Improve the Model
Task 2.1: Implement a data augmentation pipeline to improve model performance.
Task 2.2: Evaluate the trade-off between performance improvement and additional training time due to the data augmentation.
Task 2.3: Analyze the wrongly classified items and suggest improvements.
Task 3: Improve Model Generalisability
Test the model on data collected from a different source (like a different type of camera or background) to evaluate its generalizability.
Here's a breakdown based on the tasks:
Task 1: Smart Recycling using Deep Learning
Task 1.1: Define an image classification problem that could help with recycling efforts. You will have to identify the classes of recyclable items, like paper, plastic, metal, etc., and non-recyclable items.
Task 1.2: Make a plan for data collection, how you'll split the data, and how you'll evaluate the model's performance.
Task 1.3: Implement the deep learning model using the dataset you've gathered or identified.
Task 2: Analyse and Improve the Model
Task 2.1: Implement a data augmentation pipeline to improve model performance.
Task 2.2: Evaluate the trade-off between performance improvement and additional training time due to the data augmentation.
Task 2.3: Analyze the wrongly classified items and suggest improvements.
Task 3: Improve Model Generalisability
Test the model on data collected from a different source (like a different type of camera or background) to evaluate its generalizability.