Computer Vision Experiments: CIFAR-10 Classification

A comparative study of Traditional Computer Vision (HOG+SVM) vs Deep Learning (CNN) for image classification on the CIFAR-10 dataset.

AI/MLComputer Vision

# features

Key Features

Core technologies and system features.

Traditional Computer Vision

Implementation using HOG for feature extraction and Linear SVM for classification.

Deep Learning Approach

Custom 3-layer CNN architecture with TensorFlow/Keras achieving ~71.3% accuracy.

Performance Analysis

Detailed comparison across Accuracy, Precision, Recall, and F1-Score.

Reproducible Notebook

End-to-end implementation from data preprocessing to visualization in Jupyter.

# source

Project Source Code

Explore the primary logical modules.

EXPLORER

cnn_model.py

srccnn_model.py

1from tensorflow.keras import layers, models
2                        
3                        def create_cnn_model():
4                            model = models.Sequential([
5                                layers.Conv2D(32, (3, 3), activation='relu', input_shape=(32, 32, 3)),
6                                layers.BatchNormalization(),
7                                layers.MaxPooling2D((2, 2)),
8                                layers.Dropout(0.25),
9                                
10                                layers.Conv2D(64, (3, 3), activation='relu'),
11                                layers.BatchNormalization(),
12                                layers.MaxPooling2D((2, 2)),
13                                layers.Dropout(0.25),
14                                
15                                layers.Flatten(),
16                                layers.Dense(128, activation='relu'),
17                                layers.Dropout(0.5),
18                                layers.Dense(10, activation='softmax')
19                            ])
20                            return model

# simulation

Live Simulation Output

Simulated console execution.

simulation

$_▋

# repositories

Source Code

GitHub repositories for this project.

Computer Vision Experiments: CIFAR-10 Classification Repository

Access the complete source code on GitHub.