QJ PERCEPTION MODEL

py-qj-robots

QJ Robots Python SDK provides powerful machine vision perception capabilities, supporting object detection, image segmentation, attribute description, angle prediction, keypoint detection, and grasp point prediction for 2D/3D images.

Requirements

• Python 3.x
• Dependencies: requests>=2.26.0, python-dotenv>=0.19.0

Installation

pip install py-qj-robots

Configuration

The following environment variables need to be configured before using the SDK:

• QJ_APP_ID: Application ID
• QJ_APP_SECRET: Application Secret

You can click here to obtain your Application ID and Secret.

You can configure these variables in two ways:

1. Create a .env file:

QJ_APP_ID=your_app_id
QJ_APP_SECRET=your_app_secret

2. Using export command:

export QJ_APP_ID=your_app_id
export QJ_APP_SECRET=your_app_secret

Quick Start

from dotenv import load_dotenv
from py_qj_robots import Perception
import os

# Load environment variables
load_dotenv()

# Initialize Perception instance
perception = Perception()

# Perform 2D image detection
result = perception.check_image(
    image_type="2D",
    image_url="http://example.com/image.jpg",
    object_names=["bottle", "cup"]
)
print(f"Detection result: {result}")

API Documentation

Perception Class

check_image

Detect target objects in the image.

def check_image(image_type: str, image_url: str, object_names: Union[str, List[str]], depth_url: Optional[str] = None) -> Dict

Parameters:

• image_type: Image type, '2D' or '3D'
• image_url: Image URL
• object_names: Names of objects to detect, can be a string or list of strings
• depth_url: Depth image URL (required only when image_type is '3D')

split_image

Segment target objects in the image.

def split_image(image_type: str, image_url: str, object_names: Union[str, List[str]], depth_url: Optional[str] = None) -> Dict

Return value includes:

• boxes: Bounding box coordinates [x1,y1,x2,y2]
• masks: Mask image URLs and data
• croppedImagesListBbox: Cropped image URLs
• labels: Detected object labels
• scores: Confidence scores

props_describe

Get attribute descriptions of objects in the image.

def props_describe(image_type: str, image_url: str, object_names: Union[str, List[str]], questions: Union[str, List[str]], depth_url: Optional[str] = None) -> Dict

angle_prediction

Predict object angles.

def angle_prediction(image_type: str, image_url: str, object_names: Union[str, List[str]], depth_url: Optional[str] = None) -> Dict

key_point_prediction

Predict object keypoints.

def key_point_prediction(image_type: str, image_url: str, object_names: Union[str, List[str]], depth_url: Optional[str] = None) -> Dict

grab_point_prediction

Predict object grasp points.

def grab_point_prediction(image_type: str, image_url: str, object_names: Union[str, List[str]], depth_url: Optional[str] = None) -> Dict

full_perception

Perform complete perception analysis, including all features.

def full_perception(image_type: str, image_url: str, object_names: Union[str, List[str]], questions: Union[str, List[str]], depth_url: Optional[str] = None) -> Dict

Return value includes all perception results:

• angles: Angle information
• boxes: Bounding boxes
• masks: Segmentation masks
• points: Keypoints
• grasps: Grasp points
• answers: Attribute descriptions
• etc.

More Information

Visit QJ Robots Official Website for more information.