Python Client for UnrealCV

This is the official Python client for UnrealCV, a project to help computer vision researchers build virtual worlds using Unreal Engine.

Installation

pip install unrealcv

Quick Start

Launch a game with the UnrealCV plugin, then connect from Python:

from unrealcv import client
client.connect()  # Connect to the game
if not client.isconnected():
    print('UnrealCV server is not running.')
else:
    # Capture an image
    filename = client.request('vget /camera/0/lit')
    print(f'Image saved to {filename}')

    # Get camera location
    location = client.request('vget /camera/0/location')
    print(f'Camera at: {location}')

High-Level API

For more advanced usage (image processing, object manipulation, camera control), use the UnrealCv_API class:

from unrealcv import UnrealCv_API

api = UnrealCv_API(port=9000, ip='127.0.0.1', resolution=(640, 480))

# Get an RGB image as a numpy array
image = api.get_image(0, 'lit')

# Get depth map
depth = api.get_depth(0)

# List all objects in the scene
objects = api.get_objects()
for obj in objects:
    location = api.get_obj_location(obj)
    print(f'{obj}: {location}')

# Annotation helpers and asset-path spawning
api.annotate_world()
spawned = api.spawn_object_from_path('/Game/MetaHumans/Taro/BP_Taro.BP_Taro', 'SpawnedFromPath')
api.clear_world_annotation()

Note: The high-level API requires opencv-python, numpy, and pillow.

Documentation

• Full Documentation
• Command Reference
• Getting Started Tutorial

Links

• GitHub Repository
• UnrealZoo - Pre-built Binaries