Wizart Vision SDK for Python

The Wizart Vision SDK is a set of software development tools and libraries provided by Wizart.ai that enables developers to integrate the Wizart Vision API's capabilities into their applications more easily. The SDK acts as a bridge between the Vision API and the developer's application by providing a standardized set of functions and interfaces that the developer can use to access the API's functionality.

Looking for more documentation?

• Wizart Vision API Review
• Wizart Vision SDK for JavaScript

⭐️ Start using Wizart Vision API with the RapirAPI platform.

Features

Wizart Vision technology base consists of several core components that power our computer vision solutions. These include segmentation, detection, reconstruction, and analysis, each of which plays a critical role in enabling advanced visual capabilities. Below are links to learn more about each component and how they contribute to our powerful Vision API.

• Technologies for Home Interior
• Indoor Semantic Segmentation
• Indoor Object Detection
• 3D Interior Layout Reconstruction
• Indoor Scene Analysis & Classification

vision_api.mp4

Installation

pip3 install wizart-vision

Authentication

Once you received X-RapidAPI-Key, you need initialize vision client

from wizart.vision import ComputerVisionClient as vc

client = vc(
    token="Your token"
)

Usage

See notebook with examples. The client allows you to perform requests similar to those described in the documentation.

• segmentation
• detection
• reconstruction
• analysis
• interior

You will operate just with few parameters.

• resource - file system path or http link to the image
• feature - entity or surface name, available in Wizart Vision SDK

from wizart.vision import ComputerVisionClient as vc

# use this feature object for segmentation, detection, reconstruction and interior calls
vc.feature

# currently supported feature entities
vc.feature.WALL
vc.feature.CEILING
vc.feature.FLOOR
vc.feature.WINDOW

# use this object for different analysis calls
vc.analysis_types

# currently supported analysis types
vc.analysis_types.CAMERA
vc.analysis_types.IMAGE_INFO
vc.analysis_types.INTERIOR_TYPE

Segmentation

Indoor scene semantic decomposition process.

Obtaining indoor segmentation mask

mask = client.segmentation(
    resource="file system path or http link to image"
)

Segmentation by feature (i.e. by surface object)

mask = client.segmentation(
    resource="file system path or http link to image",
    feature=vc.feature.CEILING
)

To obtain only the mask contours, enable the vectorized option by setting it to True.

mask = client.segmentation(
    resource="file system path or http link to image",
    feature=vc.feature.CEILING,
    vectorized=True
)

Detection

Localize objects coordinates in the photo.

Detect all supported entities

feature = client.detection(
    resource="path to image"
)

Single entity detection, e.g. detect only the walls

feature = client.detection(
    resource="path to image",
    feature=vc.feature.WALL
)

Reconstruction

Obtain information about the 3D dimensions (real sizes) and positions of scene objects in the photo.

Reconstruct all supported entities and scene params

feature = client.reconstruction(
    resource="path to image"
)

Reconstruct a specific entity and scene params

feature = client.reconstruction(
    resource="path to image",
    feature=vc.feature.FLOOR
)

Analyze

The Analysis API includes a set of different computer vision solutions based on neural networks.

Analyse image, interior and camera

feature = client.analysis(
    resource="path to image"
)

Perform a specific type of analysis

feature = client.analysis(
    resource="path to image",
    feature=vc.analysis_types.CAMERA
)

Interior

Provides the ability to get all the data on the requested feature that we were able to extract from the uploaded interior photo.

Describe all entities

feature = client.interior(
    resource="path to image"
)

Get data for a specific entity

feature = client.interior(
    resource="path to image",
    feature=vc.feature.CEILING
)

To obtain only the mask contours, enable the vectorized option by setting it to True.

mask = client.interior(
    resource="file system path or http link to image",
    vectorized=True
)

Supported analysis types and features are listed in _client_enums.py