Scribd
Upload
13 views3 pages

Software Requirements Specification

Uploaded by

anuvagarg3
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as DOCX, PDF, TXT or read online on Scribd
13 views3 pages

Software Requirements Specification

Uploaded by

anuvagarg3
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as DOCX, PDF, TXT or read online on Scribd
You are on page 1/ 3

Software Requirements Specification (SRS)

1. Introduction
1.1 Purpose
The purpose of this document is to define the software requirements for a Skin
Cancer Detection System using Convolutional Neural Networks (CNNs).
The system aims to classify skin lesion images into different categories to assist
dermatologists and healthcare professionals in early detection and diagnosis of
skin cancer.
1.2 Scope
This project involves:
 Dataset Preparation: Using the HAM10000 dataset for training and
testing.
 Image Preprocessing: Normalizing and resizing images.
 Model Development: Implementing DenseNet121 for classification.
 Performance Evaluation: Using metrics such as accuracy, precision,
recall, and F1-score.
 Visualization: Plotting confusion matrices and training histories.
1.3 Definitions, Acronyms, and Abbreviations
 CNN: Convolutional Neural Network
 HAM10000: Human Against Machine dataset for skin lesion analysis
 DenseNet121: A deep learning architecture used for feature extraction
 Accuracy: The ratio of correct predictions to total predictions
 Precision: The ratio of correctly predicted positive observations
 Recall: The ability of the model to find all positive cases
 F1-score: The harmonic mean of precision and recall
1.4 References
 HAM10000 Dataset: Kaggle Dataset
 TensorFlow Documentation: TensorFlow
 Scikit-Learn Documentation: Scikit-Learn
2. Overall Description
2.1 Product Perspective
This project is a machine learning-based classification system that
automates skin cancer detection from medical images. The system will use deep
learning techniques to provide high accuracy in classification.
2.2 Product Functions
 Data Preprocessing
 Image Augmentation
 Training a CNN Model
 Model Evaluation & Metrics Calculation
 Visualization of Results
2.3 User Characteristics
 Doctors & Dermatologists: Use predictions as a diagnostic aid.
 Medical Researchers: Improve models for automated detection.
 Students & AI Enthusiasts: Learn from implementation.
2.4 Constraints
 Requires high computational power (GPU recommended)
 Dataset imbalance may affect classification performance
 Requires pretrained weights for DenseNet121
3. Specific Requirements
3.1 Functional Requirements
FR1: Dataset Handling
 System must download the HAM10000 dataset from Kaggle.
 System must preprocess images to 128x128 resolution.
FR2: Model Training
 System must train a DenseNet121-based CNN.
 System must use data augmentation techniques.
FR3: Model Evaluation
 System must compute accuracy, precision, recall, F1-score.
 System must plot confusion matrix and training history.
FR4: Performance Optimization
 System must fine-tune the last 30 layers of DenseNet121.
 System must use dropout layers to prevent overfitting.
3.2 Non-Functional Requirements
NFR1: Performance
 Model should achieve at least 85% accuracy on the test set.
NFR2: Usability
 System should allow easy modifications for hyperparameter tuning.
NFR3: Reliability
 System should handle missing or corrupted images gracefully.
4. Appendices
 Appendix A: Sample code for model training
 Appendix B: Instructions for dataset setup

You might also like