perceptra 1.1.0

Official Python SDK for the Perceptra Hub Computer Vision MLOps platform

Perceptra Python SDK

Official Python SDK for the Perceptra Hub Computer Vision MLOps platform.

Installation

pip install perceptra

Quick Start

import perceptra

# Initialize with your API key
client = perceptra.Perceptra(api_key="ph_live_abc123...")

# Or set PERCEPTRA_API_KEY environment variable
client = perceptra.Perceptra()

# Create a project
project = client.projects.create(
    name="Hard Hat Detection",
    project_type_id=1,
)

# Upload images
from pathlib import Path
for path in Path("./images").glob("*.jpg"):
    client.images.upload(file=path, project_id=project.project_id)

# Add annotations
client.annotations.create(
    project_id=project.project_id,
    image_id="...",
    annotation_type="bbox",
    annotation_class_id=0,
    data=[0.1, 0.2, 0.5, 0.8],
)

# Split dataset
client.projects.split_dataset(project.project_id, 0.7, 0.2, 0.1)

# Create a dataset version
version = client.versions.create(project.project_id, "v1.0")

# Create and train a model
model = client.models.create(
    project_id=project.project_id,
    name="YOLOv8 Hard Hat",
    task="object-detection",
    framework="yolo",
)

result = client.models.train(
    model_id=model.id,
    dataset_version_id=version.id,
    config={"epochs": 100, "batch_size": 16},
)

# Stream training logs
for line in client.training.stream_logs(result.training_session_id):
    print(line, end="")

Async Support

import asyncio
import perceptra

async def main():
    async with perceptra.AsyncPerceptra(api_key="ph_live_abc123...") as client:
        projects = await client.projects.list()
        for p in projects:
            print(p["name"])

asyncio.run(main())

Error Handling

from perceptra import NotFoundError, RateLimitError, AuthenticationError

try:
    project = client.projects.retrieve("nonexistent-id")
except NotFoundError:
    print("Project not found")
except RateLimitError as e:
    print(f"Rate limited. Retry after {e.retry_after}s")
except AuthenticationError:
    print("Invalid API key")

Available Resources

Resource	Description
client.projects	Project CRUD, image management, dataset splitting
client.images	Image upload, listing, bulk operations
client.annotations	Annotation CRUD, batch operations
client.models	ML model CRUD, training triggers
client.training	Training session monitoring, log streaming
client.versions	Dataset version management, export
client.organizations	Organization details, members
client.jobs	Annotation job management
client.classes	Annotation class management
client.tags	Image tag management
client.api_keys	API key management, rotation
client.storage	Storage profile management

Configuration

Parameter	Default	Environment Variable
api_key	—	PERCEPTRA_API_KEY
base_url	http://localhost:29082	PERCEPTRA_BASE_URL
timeout	30.0	—
max_retries	3	—

License

MIT