https://github.com/ornella-gigante/pytorch-visioncraft-advanced-image-transformation-studio
En este pequeño proyecto, utilizaré una imagen para hacer algunas transformaciones que se podrían usar para aumentar un conjunto de datos y mejorar el modelo utilizando varias capas y una red convolucional.
https://github.com/ornella-gigante/pytorch-visioncraft-advanced-image-transformation-studio
anaconda3 python pytorch torchvision-datasets
Last synced: 7 months ago
JSON representation
En este pequeño proyecto, utilizaré una imagen para hacer algunas transformaciones que se podrían usar para aumentar un conjunto de datos y mejorar el modelo utilizando varias capas y una red convolucional.
- Host: GitHub
- URL: https://github.com/ornella-gigante/pytorch-visioncraft-advanced-image-transformation-studio
- Owner: Ornella-Gigante
- License: gpl-3.0
- Created: 2024-02-28T18:20:08.000Z (over 1 year ago)
- Default Branch: main
- Last Pushed: 2025-02-07T10:10:12.000Z (8 months ago)
- Last Synced: 2025-03-08T12:47:57.408Z (8 months ago)
- Topics: anaconda3, python, pytorch, torchvision-datasets
- Language: Jupyter Notebook
- Homepage:
- Size: 948 KB
- Stars: 1
- Watchers: 1
- Forks: 0
- Open Issues: 0
-
Metadata Files:
- Readme: README.md
- License: LICENSE
Awesome Lists containing this project
README
# 🖼️ **Image Transformation Toolkit with PyTorch** 🛠️
*Advanced Image Processing and Visualization using PyTorch and TorchVision*
---
## 🚀 **Project Overview**
This project demonstrates **advanced image transformations** using PyTorch and TorchVision. It includes a Jupyter Notebook (`Ornella_Gigante_Lab4_Sprint2.ipynb`) that applies various photometric and geometric adjustments to images, such as brightness, contrast, gamma correction, and hue/saturation shifts. The toolkit also visualizes the effects of these transformations, enabling users to compare original and modified images side-by-side.
**Key Use Cases**:
- Data augmentation for deep learning models.
- Image preprocessing for computer vision tasks.
- Educational tool for understanding transformation effects.
---
## 🌟 **Core Features**
| Transformation | Description | Example Use Case |
|--------------------------|-----------------------------------------------------------------------------|-----------------------------------|
| **Brightness Adjustment**| Modifies image illumination (`adjust_brightness`). | Simulate low-light conditions. |
| **Contrast Enhancement** | Adjusts difference between light and dark areas (`adjust_contrast`). | Improve feature visibility. |
| **Gamma Correction** 🌓 | Non-linear adjustment for perceptual brightness (`adjust_gamma`). | Correct underexposed images. |
| **Hue/Saturation** 🎨 | Alters color tones and intensity (`adjust_hue`, `adjust_saturation`). | Artistic style transfer. |
| **Sharpness Optimization** 🔍 | Enhances edge clarity (`adjust_sharpness`). | Improve OCR readability. |
| **Visual Comparison** | Side-by-side display of original vs. transformed images. | Quality assessment. |
---
## 🛠️ **Tech Stack**
- **PyTorch**: Tensor operations and GPU acceleration.
- **TorchVision**: Prebuilt transformations and image utilities.
- **PIL/Pillow**: Image loading and preprocessing.
- **Matplotlib**: Visualization of results.
- **Jupyter Notebook**: Interactive experimentation.
---
## 📥 **Installation**
1. **Clone Repository**:
```bash
git clone https://github.com/yourusername/image-transformation-toolkit.git
cd image-transformation-toolkit
```
2. **Install Dependencies**:
```bash
pip install torch torchvision pillow matplotlib
```
3. **Launch Jupyter Notebook**:
```bash
jupyter notebook Ornella_Gigante_Lab4_Sprint2.ipynb
```
---
## 🖌️ **Code Examples**
### **1. Image Loading & Tensor Conversion**
```python
from PIL import Image
from torchvision.transforms import functional as F
img = Image.open('path/to/image.jpg')
img_tensor = F.to_tensor(img) # Convert to PyTorch tensor
```
### **2. Applying Transformations**
```python
# Adjust brightness (50% reduction)
brightness_adjusted = F.adjust_brightness(img_tensor, brightness_factor=0.5)
# Increase contrast by 150%
contrast_enhanced = F.adjust_contrast(img_tensor, contrast_factor=1.5)
```
### **3. Visualization**
```python
import matplotlib.pyplot as plt
fig, axes = plt.subplots(1, 2, figsize=(12, 6))
axes[0].imshow(F.to_pil_image(img_tensor))
axes[0].set_title("Original Image")
axes[1].imshow(F.to_pil_image(brightness_adjusted))
axes[1].set_title("Brightness Adjusted")
plt.show()
```
---
## 📊 **Transformation Matrix**
| Parameter | Typical Range | Effect Visualization |
|-----------------------|---------------|-----------------------|
| `brightness_factor` | 0.0 (black) to 1.0+ | Brightness |
| `contrast_factor` | 0.0 (gray) to 1.0+ | Contrast |
| `hue_factor` | -0.5 to 0.5 | Hue |
---
## 📜 **License**
MIT License
---
👩💻 **Author**: Ornella Gigante
*"Transform images like a pro with PyTorch's powerful vision toolkit!"* 🚀