Scribd
Upload
6 views2 pages

Leukemia Detection with Deep Learning

This document outlines an assessment focused on image classification for leukemia detection using deep learning models. It details the dataset, which includes images of blood cells categorized into six types, and provides a step-by-step guide for model training, evaluation, and comparison. The deliverables include code implementation, evaluation results, and a recommendation for the best performing model.

Uploaded by

shahriarsajal56
Copyright
© All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
6 views2 pages

Leukemia Detection with Deep Learning

This document outlines an assessment focused on image classification for leukemia detection using deep learning models. It details the dataset, which includes images of blood cells categorized into six types, and provides a step-by-step guide for model training, evaluation, and comparison. The deliverables include code implementation, evaluation results, and a recommendation for the best performing model.

Uploaded by

shahriarsajal56
Copyright
© All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd

Computer Vision Assessment

1 Assessment Description
This assessment focuses on image classification using deep learning models for leukemia type detection. You
will use different deep learning architectures to train models and evaluate their performance. Based on the
results, you should suggest the best model for the dataset.

2 Dataset Description
The dataset contains images of blood cells categorized into six classes:
1. MM (Multiple Myeloma)

2. Healthy
3. CML (Chronic Myeloid Leukemia)
4. CLL (Chronic Lymphocytic Leukemia)
5. AML (Acute Myeloid Leukemia)

6. ALL (Acute Lymphoblastic Leukemia)

3 Usage
Split the dataset into training (60%), validation (20%), and test (20%) sets.

4 Step-by-step guide
1. Download the dataset and upload it to your preferred development environment (e.g., Google Colab,
local machine).

2. Preprocess the image data.


3. Split the dataset into training (60%), validation (20%), and test (20%) sets.
4. Build and train the following deep learning models:

i) Convolutional Neural Network (CNN) from scratch


ii) Transfer Learning with pre-trained models
iii) Any other architecture of your choice
5. Implement techniques to prevent overfitting (e.g., dropout, early stopping, data augmentation).

6. Perform hyperparameter tuning to optimize each model’s performance.


7. Use the validation set to fine-tune the models and prevent overfitting.

1
8. Evaluate the models using the test set.
9. Use appropriate evaluation metrics such as accuracy, precision, recall, F1-score, and confusion matrix.
10. Visualize the evaluation results using appropriate plots (e.g., accuracy curves, confusion matrices,
sample predictions).

11. Compare the performance of different models and suggest the BEST architecture based on accuracy
and other relevant metrics.

5 Deliverables
1. Code implementation of the entire pipeline.
2. Evaluation results and visualizations for each model.
3. A final recommendation on the best performing model with justification.

6 Note
Ensure that you comment your code thoroughly and follow best practices for clean and efficient implemen-
tation. Consider the trade-offs between model complexity, training time, and performance in your final
recommendation.

You might also like