deeplearning-lab-ktu

Deep Learning Lab Activities

A comprehensive collection of educational deep learning programs demonstrating fundamental concepts in computer vision, natural language processing, and sequence modeling.

🚀 Quick Start - Web Interface (Recommended)

The easiest way to run and visualize the labs is through the web interface:

1. Install Dependencies

pip install -r requirements.txt
cd web_interface
pip install -r requirements.txt

2. Start the Web Server

cd web_interface
python app.py

3. Open in Browser

Navigate to http://localhost:5000 in your web browser.

4. Run Labs

Browse all 10 labs from the dashboard
Click on any lab card to open its interface
Click “Run Lab” to execute
View real-time console output
See generated visualizations automatically

Features:

Interactive web dashboard with Full Labs, Lite Labs, and Documentation sections
Real-time console output display
Automatic visualization of results
Run/Stop controls for each lab
Built-in documentation viewer
Modern dark theme with responsive design

For detailed web interface documentation, see web_interface/README.md

🐍 Using Virtual Environment (Recommended)

It’s highly recommended to use a virtual environment to avoid dependency conflicts:

Option 1: Using venv (Python built-in)

# Create virtual environment
python -m venv venv

# Activate virtual environment
# On macOS/Linux:
source venv/bin/activate
# On Windows:
venv\Scripts\activate

# Install dependencies
pip install -r requirements.txt
cd web_interface
pip install -r requirements.txt
cd ..

# Run web interface
cd web_interface
python app.py

Option 2: Using conda

# Create conda environment
conda create -n dl-lab python=3.9

# Activate environment
conda activate dl-lab

# Install dependencies
pip install -r requirements.txt
cd web_interface
pip install -r requirements.txt
cd ..

# Run web interface
cd web_interface
python app.py

Deactivating Virtual Environment

When you’re done:

# For venv:
deactivate

# For conda:
conda deactivate

Benefits of Virtual Environment:

Isolated dependencies from system Python
Prevents version conflicts with other projects
Easy to reproduce exact environment
Clean uninstall by simply deleting the environment

📚 Overview

This repository contains 10 self-contained lab activities designed for academic learning. Each lab includes:

Well-commented, production-quality code
Clear learning objectives
Sample datasets or data generation
Training and evaluation pipelines
Visualization of results
Execution time under 5 minutes on standard hardware

🗂️ Lab Activities

Lab 1: Basic Image Processing Operations

Directory: labs_full/lab_01_image_processing/ | Lite: labs_lite/lab_01_image_processing/

Histogram equalization
Thresholding techniques
Edge detection (Sobel, Canny)
Data augmentation
Morphological operations

Lab 2: CIFAR-10 Classifiers

Directory: labs_full/lab_02_cifar10_classifiers/ | Lite: labs_lite/lab_02_cifar10_classifiers/

KNN classifier implementation
3-layer neural network classifier
SVM/Softmax classifier comparison

Lab 3: Batch Normalization and Dropout Study

Directory: labs_full/lab_03_batchnorm_dropout/ | Lite: labs_lite/lab_03_batchnorm_dropout/

Effect of batch normalization on training
Impact of dropout on overfitting
Comparative analysis with visualizations

Lab 4: Image Labeling Tools Demonstration

Directory: labs_full/lab_04_labeling_tools/ | Lite: labs_lite/lab_04_labeling_tools/

Synthetic image generation with annotations
Format conversions (COCO, YOLO, Pascal VOC)
Annotation visualization
Statistics generation

Lab 5: Image Segmentation with UNet

Directory: labs_full/lab_05_segmentation/ | Lite: labs_lite/lab_05_segmentation/

UNet architecture implementation
Comparison with simple CNN baseline
Evaluation metrics (IoU, Dice coefficient)
Synthetic segmentation dataset

Lab 6: Object Detection with YOLO

Directory: labs_full/lab_06_object_detection/ | Lite: labs_lite/lab_06_object_detection/

YOLO-style object detector
Bounding box prediction and classification
Non-Maximum Suppression (NMS)
Detection visualization

Lab 7: Image Captioning with RNN/LSTM

Directory: labs_full/lab_07_image_captioning/ | Lite: labs_lite/lab_07_image_captioning/

CNN encoder + RNN/LSTM decoder
Comparison between Vanilla RNN and LSTM
Caption generation
Synthetic image-caption pairs

Lab 8: Chatbot with Bi-directional LSTM

Directory: labs_full/lab_08_chatbot/ | Lite: labs_lite/lab_08_chatbot/

Bi-directional LSTM for intent classification
Comparison with unidirectional LSTM
Response generation
Interactive demo

Lab 9: Time Series Forecasting with LSTM

Directory: labs_full/lab_09_time_series/ | Lite: labs_lite/lab_09_time_series/

LSTM and GRU for forecasting
Multi-step ahead prediction
Comparison with baseline
Evaluation metrics (MSE, MAE, RMSE, MAPE)

Lab 10: Sequence to Sequence Learning

Directory: labs_full/lab_10_seq2seq/ | Lite: labs_lite/lab_10_seq2seq/

Encoder-decoder architecture
Attention mechanism
Machine translation (synthetic language)
Beam search decoding

💻 Alternative: Command Line

If you prefer running labs directly from the command line:

Prerequisites

Python 3.8 or higher
pip package manager
(Optional) CUDA-capable GPU for faster training

Installation

Clone or download this repository
Install dependencies:
```
pip install -r requirements.txt
```

Running Individual Labs

Each lab is self-contained with an executable Python program. Navigate to the lab directory and run:

# Lab 1: Image Processing
cd labs_full/lab_01_image_processing
python image_processing.py

# Lab 2: CIFAR-10 Classifiers
cd labs_full/lab_02_cifar10_classifiers
python cifar10_classifiers.py

# Lab 3: Batch Normalization & Dropout
cd labs_full/lab_03_batchnorm_dropout
python batchnorm_dropout_study.py

# Lab 4: Labeling Tools Demo
cd labs_full/lab_04_labeling_tools
python labeling_demo.py

# Lab 5: Image Segmentation
cd labs_full/lab_05_segmentation
python segmentation_demo.py

# Lab 6: Object Detection
cd labs_full/lab_06_object_detection
python object_detection_demo.py

# Lab 7: Image Captioning
cd labs_full/lab_07_image_captioning
python image_captioning_demo.py

# Lab 8: Chatbot
cd labs_full/lab_08_chatbot
python chatbot_demo.py

# Lab 9: Time Series Forecasting
cd labs_full/lab_09_time_series
python time_series_demo.py

# Lab 10: Sequence to Sequence
cd labs_full/lab_10_seq2seq
python seq2seq_demo.py

Or run all labs sequentially:

cd labs_full
bash run_all_labs.sh

For lightweight versions with faster execution:

cd labs_lite
bash run_all_labs.sh

Note: Each program creates an output/ directory in its lab folder with visualizations and results.

📦 Dependencies

Core libraries used across labs:

PyTorch: Deep learning framework
TensorFlow/Keras: Alternative framework for some labs
OpenCV: Image processing
NumPy: Numerical computations
Matplotlib/Seaborn: Visualization
scikit-learn: Classical ML algorithms
Pillow: Image handling
pandas: Data manipulation

See requirements.txt for complete list with versions.

📊 Expected Execution Times

All labs are designed to complete within 5 minutes on standard hardware (CPU):

Lab	Program	Execution Time	Notes
Lab 1	`image_processing.py`	< 30 seconds	Image processing operations
Lab 2	`cifar10_classifiers.py`	2-3 minutes	KNN, SVM, 3-layer NN training
Lab 3	`batchnorm_dropout_study.py`	2-3 minutes	4 model variants comparison
Lab 4	`labeling_demo.py`	< 30 seconds	Synthetic data generation
Lab 5	`segmentation_demo.py`	3-4 minutes	UNet + baseline training
Lab 6	`object_detection_demo.py`	3-4 minutes	YOLO-style detector training
Lab 7	`image_captioning_demo.py`	3-4 minutes	RNN + LSTM training
Lab 8	`chatbot_demo.py`	2-3 minutes	BiLSTM + LSTM training
Lab 9	`time_series_demo.py`	2-3 minutes	LSTM + GRU forecasting
Lab 10	`seq2seq_demo.py`	3-4 minutes	Seq2Seq + Attention training

Total time for all labs: ~25-30 minutes

With GPU: Execution times can be 2-5x faster depending on GPU model.

Lite versions: Execute 2-3x faster with reduced dataset sizes and epochs.

🎓 Learning Objectives

By completing these labs, you will:

Understand fundamental image processing techniques
Implement and compare different classification approaches
Master regularization techniques in neural networks
Learn data annotation workflows for computer vision
Build semantic and instance segmentation models
Implement modern object detection architectures
Create image captioning systems with RNNs and LSTMs
Develop conversational AI with bi-directional LSTMs
Apply deep learning to time series problems
Understand sequence-to-sequence architectures

📝 Repository Structure

dl-lab/
├── web_interface/         # Web-based interface (recommended)
│   ├── app.py            # Flask application
│   ├── templates/        # HTML templates
│   └── requirements.txt  # Web interface dependencies
├── labs_full/            # Full-featured lab programs
│   └── lab_XX_name/
│       ├── README.md     # Lab-specific instructions
│       ├── program.py    # Main executable
│       └── output/       # Generated results (created on run)
├── labs_lite/            # Lightweight versions (faster execution)
│   └── lab_XX_name/
│       ├── program_lite.py
│       └── output/
├── docs/                 # Additional documentation
└── requirements.txt      # Main dependencies

All programs:

Are fully self-contained and executable
Generate synthetic data or download datasets as needed
Create visualizations and save results in output/ directory
Include comprehensive comments and documentation
Print progress and results to console
Available in both full and lite versions

🔧 Troubleshooting

Common Issues

Import Errors:

pip install --upgrade -r requirements.txt

CUDA Out of Memory:

Reduce batch size in the code
Use CPU mode by setting device = 'cpu'

Slow Execution:

Ensure you’re using GPU if available
Reduce dataset size or number of epochs
Check system resources

📚 Documentation

For more detailed information, see:

Web Interface Guide - Complete web interface documentation
Installation Guide - Detailed setup instructions
Dataset Setup Guide - Dataset download and management
Getting Started - Beginner’s guide
Advanced Tab Guide - Advanced features

📚 Additional Resources

🤝 Contributing

This is an educational repository. Feel free to:

Report issues or bugs
Suggest improvements
Add more examples or variations

📄 License

This project is created for educational purposes. Feel free to use and modify for learning.

✨ Acknowledgments

These labs are designed for academic learning and incorporate best practices from:

Stanford CS231n: Convolutional Neural Networks
Stanford CS224n: Natural Language Processing
Fast.ai courses
PyTorch and TensorFlow tutorials

Note: All programs are designed to be self-contained and educational. They prioritize clarity and learning over production optimization.

Last Updated: March 2026