depthai

What it is and what it does

DepthAI is a Python wrapper around the DepthAI C++ core library, providing direct access to Luxonis OAK depth cameras and spatial AI accelerators. It exposes a high-level API for configuring camera pipelines, capturing depth and RGB frames, and running on-device neural network inference. The library depends only on numpy and is distributed as prebuilt wheels for modern Python versions across Windows, macOS, and Linux.

Typical workflows involve creating a pipeline (defining camera nodes, neural networks, and data flows), connecting it to a device, and then streaming frames or inference results back to Python. The package is designed for real-time 3D vision tasks where depth sensing and spatial awareness are primary requirements. Building from source is supported but adds significant complexity due to C++ compilation and external dependencies like libusb.

Use it for:

• Real-time depth sensing and 3D reconstruction from OAK stereo cameras in robotics or autonomous systems.
• On-device spatial AI inference (object detection, pose estimation) without sending frames to a remote server.
• Prototyping computer vision pipelines that combine depth, RGB, and neural network outputs on embedded or edge hardware.
• Building augmented reality or spatial mapping applications that require low-latency depth data.
• Integrating OAK cameras into Python-based research or educational projects in computer vision.

Worth the install?

AI-flagged interpretation of the facts on this page — verify before relying

Python bindings for the DepthAI C++ library, enabling programmatic control of OAK (OmniDirectional Awareness Kit) depth cameras and spatial AI hardware.

Yes, if you own or plan to use an OAK camera or compatible Luxonis depth hardware. The library is actively maintained, has no known vulnerabilities, and offers permissive licensing. Install friction is moderate due to compiled bindings and system dependencies, but prebuilt wheels mitigate most friction for standard platforms. Not worth installing without the hardware.

Install

depthai on PyPI

pip install depthai

uv add depthai

poetry add depthai

Installing depthai

Before you install

Medium install friction due to compiled C++ bindings and platform-specific wheels. Prebuilt wheels available for Python 3.9–3.14 on macOS (arm64, x86_64), Linux (aarch64, x86_64), and Windows (amd64). Building from source requires cmake ≥ 3.4, a C++14 compiler, and depthai-core dependencies (including libusb); common issues include missing system libraries and Hunter cache conflicts.

License in practice

MIT license (permissive) places no restrictions on use, modification, or distribution in proprietary or open-source projects.

Quickstart

pip install depthai
import depthai as dai
pipeline = dai.Pipeline()
cam = pipeline.createColorCamera()
device = dai.Device(pipeline)

Requires a connected OAK camera or compatible depth hardware; building from source requires cmake ≥ 3.4 and a C++14 compiler.

Verify before relying

• Specific OAK camera models and hardware versions supported by version 3.8.0
• Performance characteristics (latency, throughput) for typical depth/spatial AI workloads
• Whether prebuilt wheels include all optional features or if some require source build

Package facts

Evidence: depthai-3.8.0-cp39.cp310.cp311.cp312.cp313.cp314-cp39.cp310.cp311.cp312.cp313.cp314-macosx_11_0_arm64.whl; depthai-3.8.0-cp39.cp310.cp311.cp312.cp313.cp314-cp39.cp310.cp311.cp312.cp313.cp314-macosx_11_0_x86_64.whl; depthai-3.8.0-cp39.cp310.cp311.cp312.cp313.cp314-cp39.cp310.cp311.cp312.cp313.cp314-manylinux_2_28_aarch64.whl; depthai-3.8.0-cp39.cp310.cp311.cp312.cp313.cp314-cp39.cp310.cp311.cp312.cp313.cp314-manylinux_2_28_x86_64.whl; depthai-3.8.0-cp39.cp310.cp311.cp312.cp313.cp314-cp39.cp310.cp311.cp312.cp313.cp314-win_amd64.whl

Tags

More Software Development packages