Skip to content

Depthlib

Depthlib is a Python library for depth estimation using stereo vision and monocular deep learning models.

The library provides simple, high-level APIs that abstract away internal image processing and stereo matching details, making it easy to estimate depth from images and video streams.

Features

  • Stereo Depth Estimation - Compute depth maps from rectified stereo image pairs using Semi-Global Block Matching (SGBM)
  • Live Video Depth - Real-time depth estimation from synchronized stereo video streams or cameras
  • Monocular Depth - Single-image depth estimation using pre-trained deep learning models (Depth Anything V2)
  • Configurable Pipeline - Fine-tune SGBM parameters for optimal results in your scene
  • Visualization Tools - Built-in functions for displaying disparity and depth maps

Supported Modes

Mode Class Description
Stereo Images StereoDepthEstimator Depth from a single pair of stereo images
Stereo Video StereoDepthEstimatorVideo Real-time depth from video streams/cameras
Monocular MonocularDepthEstimator Depth from a single RGB image using AI

Quick Start

Installation

pip install depthlib

Development Setup

# Clone the repository
git clone https://github.com/your-username/depthlib.git

# Create virtual environment
python -m venv venv

# Activate (Windows)
venv\Scripts\activate

# Activate (macOS/Linux)
source venv/bin/activate

# Install dependencies
pip install -r requirements.txt

Basic Usage

import depthlib

# Stereo depth estimation
estimator = depthlib.StereoDepthEstimator(
    left_source='left.png',
    right_source='right.png',
    downscale_factor=0.5
)
estimator.configure_sgbm(
    num_disp=128,
    focal_length=3997.684,
    baseline=0.193
)
disparity, depth = estimator.estimate_depth()
estimator.visualize_results()

Requirements

  • Python 3.8+
  • OpenCV
  • NumPy
  • Matplotlib
  • PyTorch (for monocular depth)
  • Transformers (for monocular depth)

Next Steps