Full Academic Deep Learning / Machine Learning Project – Cardiomegaly Detection

Job ID: 40041669

Budget: $10 – $30 USD

I am looking for an experienced Deep Learning / Machine Learning professional to complete a full academic-style project.

I will provide:

Dataset (train, validation, test images)

Example baseline code (starter CNN + DenseNet121 example)

Assignment brief and marking criteria

Your job is to complete the full project to a high academic standard.

Component 1 — Coursework (70%)

Task

Build and compare two models for Cardiomegaly detection in chest X-ray images:

Baseline CNN (custom architecture)

DenseNet121 (transfer learning)

What I Will Give You

Full dataset (224×224 images)

Starter code for both models

Example paper / similar study

Your Responsibilities

Model Development

Train/validate/test both models

Tune hyperparameters

Address issues such as:

Overfitting

Class imbalance

Generate evaluation metrics:

Accuracy

Precision

Recall

F1 Score

AUC-ROC

Confusion Matrix

Produce clear graphs and tables to present results

Written Report (~3000 words)

The report must include:

Introduction

Literature Review

Methodology

Experiments & Results

Discussion

Conclusion

Appendix with code snippets & figures

Referencing style: Harvard

Output

PDF report

ZIP file with all scripts and relevant files

Component 2 — Presentation (30%)

You will also prepare:

✔ PowerPoint presentation (max 10 slides)

✔ ~1000 words of speaker notes

✔ Harvard reference list

You may choose one topic from a list (e.g., CNNs, Image Processing, Multimodal AI, Ethics in AI, etc.).

Required Skills

TensorFlow or PyTorch

CNNs & transfer learning

Experience with DenseNet121

(Bonus) Medical imaging understanding

Strong academic writing skills

Ability to generate clear visual results (graphs, plots, tables)

Deliverables

3000-word PDF report

Baseline CNN script (.py or .ipynb)

DenseNet121 script (.py or .ipynb)

Training logs, ROC curves, confusion matrix, and other graphs

PowerPoint slides (up to 10)

Speaker notes (~1000 words)

Full Harvard reference list

Important – Pricing & Policy

Your bid must reflect your final price for completing all deliverables listed above.

Requests to increase the price after the project is awarded will not be accepted.

If you attempt to charge more than your original bid or pressure for extra payment beyond what is agreed on the platform, this will be reported to Freelancer as a violation of platform rules.

Only bid if you fully understand the scope and are confident you can deliver everything at the price you submit.