Software Specification

1 Introduction

This document defines the specification of the RecycleIT software. This software is a part of a larger project to increase the efficiency and accuracy of the recycling process. In the future, we need to replace human workers with robots. Also, this software helps sustainability by improving recycling cycle. This software is the first step towards the development of recycling robots.

2 Purpose

The project has the following goals:

• 2.1 We want to reduce the costs of the recycling process.
  • 2.1.1 At the end of this project we should achieve a working prototype.
• 2.2 We need to increase the accuracy of the recycling process by having a reliable automated process.
• 2.3 We need a fast classification process due to the high load of recycled material daily.
  • 2.3.1 The software will be used in a recycling factory.
  • 2.3.2 The stream of images (video) are our inputs.

3 Features

The following features are essential in the software:

Classification

• 3.1 The software should be able to separate various materials such as Aluminum cans, and plastic.
  • 3.1.1 We need to classify different colors of plastic.
  • 3.1.2 We need to distinguish different types of plastic.
  • 3.1.3 We need to separate other unexpected objects.
  • 3.1.4 We need to clarify what "others" means.

Accuracy

• 3.2 The operational accuracy of the classification process should not be lower than 95%.
  • 3.2.1 The minimum accuracy for separation of PET will be 98%.

Performance

• 3.3 We need to classify objects at least with the rate of 10 Hz.

Dataset collection and labeling

• 3.4.1 we need to create Automatic labeling tool to facilitate dataset creation.
• 3.4.2 We have to check the correctness of labeling with statistical methods. Maybe we need to have two steps of labeling verification.
• 3.4.3 An automatic image extractor based on web scraping will enrich the dataset.
• 3.4.4 The dataset will be augmented to increase the size of dataset.
  • 3.4.4.1 [optional] Various deformation types can be labeled.

Preprocessing on images

• 3.5.1 The format, resolution and size of the input images need to be standardize by the software.
• 3.5.2 The software needs to unify the background of images.
• 3.5.3 The software needs to have an object detection tool.

4 Limitations and restrictions

The project has the following limitations:

Time

• 4.1 We only have 12 week to have the first prototype.

Software

• 4.2 The software should be easily extensible and scalable.
• 4.3 The dataset will be imbalanced. Therefore the classification precision for various classes will be different.

Hardware

• 4.4 In the production line we have a CCD sensor with moderate resolution.
• 4.5 The inference hardware is an Nvidia Jetson board.