Scribd
Upload
15 views

DLIR Oct 2024 Assignment

Uploaded by

Joel Lim
Copyright
© © All Rights Reserved
Available Formats
Download as DOCX, PDF, TXT or read online on Scribd
15 views

DLIR Oct 2024 Assignment

Uploaded by

Joel Lim
Copyright
© © All Rights Reserved
Available Formats
Download as DOCX, PDF, TXT or read online on Scribd
You are on page 1/ 4

Official (Closed) - Non Sensitive

School of InfoComm Technology

Deep Learning Assignment


Specialist Diploma in applied Generative AI
(SDGAI)
Oct 2024 Semester

ASSIGNMENT
(60% of DLIR Module)

28th Oct 2024 – 4th Dec 2024

Submission Deadline:
Presentation: 2nd OR 3rd Dec 2024 (Mon/Tue)
Report: 4th Dec 2024 (Wednesday), 11:59PM

Student Name :

Student ID :

Penalty for late submission:


10% of the marks will be deducted every calendar day after the deadline.
NO submission will be accepted after 14th December 2024 (Saturday), 11:59PM.
Official (Closed) - Non Sensitive

DLIR Assignment

1 Assignment Specifications
1.1 Objective

Build an image classification model to recognize and classify 10 different types of food.

1.2 Dataset

Download the images / dataset from POLITEMall.


 food_list.txt: the list of 10 different types of food
 food.zip: the food images split into:
o Training (750 images per food)
o Validation (200 images per food)
o Testing samples (50 images per food)

1.3 Suggested Tasks

Assignment_[your_name].ipynb Jupyter Notebook file is provided for you to write the


codes. It is recommended to tackle this problem following the below structure.

Step 1 – Load Data


 Load the Train, Validation and Test images into Jupyter Notebook
 Resize all the images to 150 x 150 (recommended)
- If it takes too long for your computer to run, you can downsize the images to
100 x100, 50 x 50 and etc.

Step 2 – Develop the Image Classification Models


- You are required to develop at least TWO different models:
 One Model trained from scratch using conv2D & dense layers
 One Model utilizing pre-trained models

 For each model, you are suggested to follow the universal machine learning workflow
to develop the model and improve the model performance, i.e.
- Start with a baseline model
- Scale up the model until overfitting occurs

DLIR Oct 2024 - Page 2 - Last Update: 21/10/2024


Official (Closed) - Non Sensitive

DLIR Assignment

- Regularize the model accordingly

 Analyze the Model Performance and Tuning the Model Hyperparameters during
training phase
- Remember to record model performance curves for report writing

Step 3 – Evaluate the models using Test Images


 Evaluate the developed models using Test Images
 Compare the model performance during testing phase
 Recommend the best model

Step 4 – Use the Best Model to make prediction


 Based on your assigned food types, download at least THREE food images from
internet
 Feed the images into your Best Model and see whether the model can classify the
food images correctly

1.4 Report Format & Content Guidelines


Write an INDIVIDUAL report with the following sections (see Table below).

Suggested content description and word count are provided for each section. You are free to
include other relevant information you deem necessary in the sections.

(Note: For a page with 1-inch margins, 12-point Arial font, and minimal spacing elements, a good rule
of thumb is 500 words for a single-spaced page)

Suggested Report Sections & Content Guidelines Word Count


1 Table of Contents NA
.
2 Overview
.  Describe the problem, the objective and the approach. Min: 300 words
Max: 1000 words

3 Data Preprocessing and Data Loading


.  Describe how you preprocess the data and load data into Jupyter Min: 200 words
Notebook
Max: 500 words

4 Develop the Image Classification Models (at least TWO)


.  Describe how you build & train the models, tuning model Min: 1000 words
hyperparameters
Max: 2000 words
 Analyze the model performance during training phase

5 Evaluate models using Test images Min: 300 words


.  Compare the models’ performance during testing phase Max: 1000 words
 Recommend the best model and explain why

DLIR Oct 2024 - Page 3 - Last Update: 21/10/2024


Official (Closed) - Non Sensitive

DLIR Assignment

6 Use the Best Model to perform classification


.  Explain how to apply the model on real life images Min: 200 words
 Explain and analyze the model prediction Max: 500 words

7 Summary
.  Summarize your model performance and provide suggestions for Min: 100 words
further improvements Max: 500 words

2 Presentation and Demonstration

Each student will be given 10 minutes to present and demo their work in Week 8 (Monday
2nd December or Tuesday 3rd December 2024, 6-10.30pm).

The presentation will be done remotely through MS Teams. The tutor will release the
detailed schedule later.

3 Deliverables
For this assignment, you must submit all the following items into POLITEMall in a single Zip
file by Wednesday 4th December 2024, 11:59pm:
1. A set of Final Presentation Slides PowerPoint slide
2. A softcopy Final Report words document
3. The completed “Assignment_[your_name].ipynb” Jupyter Notebook File
4. You Do Not need to include any .keras files

4 Grading Criteria
Component
Grading Criteria
Weightage

a) Quality of work
b) Flow of presentation based on content guidelines
Video Recorded
(see section 1.4) 50%
Presentation
c) Quality of presentation slides
d) Presentation and articulation skills

a) Quality of work
b) Completeness of report based on suggested report
Final Report sections and content guidelines (see section 1.4) 50%
c) Clarity of report, quality of analysis and discussions
d) Use of proper visual aids and proper grammar

DLIR Oct 2024 - Page 4 - Last Update: 21/10/2024

You might also like