🎯 Underwater Depth Estimation App

An advanced application for estimating and visualizing depth in underwater images and videos using state-of-the-art AI models. Built with Gradio, PyTorch, and the Depth Anything model family, optimized for underwater environments.

🌟 Features

• Multi-Format Support: Process both underwater images and videos

• Interactive Interface: User-friendly Gradio web interface

• Underwater-Specific Enhancements: Contrast correction for murky water and handling of unique lighting conditions

• Multiple Visualization Options: Various colormaps optimized for underwater depth visualization

• Analysis Tools: Depth distribution histogram for detailed analysis

• Model Options: Choice between different model sizes tailored for underwater use

• Real-Time Processing: Live preview of results

  

🔧 Installation

1. Clone the repository:

git clone https://github.com/yourusername/underwater-depth-estimation-app.git
cd underwater-depth-estimation-app

2. Create a virtual environment (recommended):

python -m venv venv
source venv/bin/activate  # On Windows use: venv\Scripts\activate

3. Install required packages:

pip install -r requirements.txt

📦 Requirements

torch
torchvision
transformers
opencv-python
pillow
gradio
tqdm
matplotlib
numpy

🚀 Usage

1. Start the Gradio app:

python app.py

2. Open your browser and navigate to the provided URL (typically http://localhost:7860)

3. Choose between image or video processing:

   • Upload an underwater image or video
   • Select your preferred colormap and model
   • Enable/disable histogram and underwater contrast enhancement
   • Click "Process" to generate the depth visualization

🎨 Available Colormaps

• INFERNO (recommended for underwater)
• JET
• VIRIDIS
• PLASMA
• MAGMA
• HOT

🤖 Available Models

• Depth Anything Small: Faster processing, smaller memory footprint
• Depth Anything Base: Higher accuracy, larger model size

📊 Output Examples

The app generates:

• Depth maps with underwater-specific color visualization
• Depth distribution histograms (for images)
• Processed videos with underwater depth visualization

🔧 Troubleshooting

Common Issues and Solutions

1. Murky or Low-Visibility Water:

   • Enable contrast enhancement
   • Use the "Base" model for better depth estimation

2. CUDA Out of Memory:

   • Reduce input image/video resolution
   • Use the "Small" model instead of "Base"
   • Free up GPU memory

3. Slow Processing:

   • Ensure CUDA is available
   • Check GPU usage
   • Consider reducing input resolution

4. Video Codec Issues:

   • Install required codecs
   • Try converting video to MP4 format
   • Check OpenCV installation