Reference no: EM133870539 , Length: word count:1500
Machine Learning Applications
Assessment - Design a Fruit Classification System
Task
Design a fruit classification system to identify the fruit name shown in the image.
Assessment Description
Your organisation is helping local farmers with sorting their produce based on the fruit labels on the boxes. It will aid farmers to stack similar products in the same pallet in a lesser amount of time.
The starting point for this project is to create a fruit classification system based on the supplied image. As a computer vision expert your team lead asked you to create a fruit classification system.
Learning Objective 1: Explore programming functions to source, store and prepare data for machine learning applications.
Learning Objective 2: Design algorithmic models for the application of machine learning in information technology.
Learning Objective 3: Create advanced insights of strategic organisational value with the aid of machine learning.
Data
A fruits dataset is curated by pre-processing the Kaggle Fruit classification dataset ( https://www.kaggle.com/datasets/sshikamaru/fruit-recognition/data) and is provided to you in MyKBS. You are encouraged to explore the original source.
The original dataset is pre-processed and is provided in 2 folders - train and test. MyKBS provides you these folders each containing 14 folders each with the respective fruit images.
You are required to train a fruit classification system using the train images. And test the system using the test images.
Problem Statement
As an individual, you are required to download the data, i.e., train and test folders from MyKBS. You must build a fruit classification system to identify similar fruits based on the supplied image. You should systematically approach the problem by addressing the below tasks:
Load the data, inspect, and pre-process it to fit your requirements. As a pre-processing step split the train data into train and cross-validation data.
Design a fruit classification system using Convolutional neural network (CNN).
Tune at least 2 hyper parameters of the base CNN. And report the best hyper parameters to use.
Write an analytical report to elaborate the approach and the performance using relevant metric(s) of the CNNs for a non-technical reader. Your report should contain the abstract, introduction, methodology and a conclusion section. The referencing is done in accordance with Kaplan Harvard Referencing style.