API Capabilities
The Matroid API allows you to build, test and deploy your computer vision detectors programmatically. Choose one of Matroid’s pre-made detectors, create a custom detector with your own examples, or import your own TensorFlow model into Matroid. Use your detector via the Matroid API to classify images, locate objects in videos, search existing collections of media, and monitor live video streams for objects and events of interest.
Quickstart with Matroid’s API Integration
CLASSIFY
CLASSIFY
MONITOR
UPLOAD
SEARCH
CLASSIFY
MONITOR
UPLOAD
SEARCH
CURL
JAVASCRIPT
PYTHON
# Classify an image from a URL
curl -X POST "https://app.matroid.com/api/v1/detectors/detector-id/classify_image" \
-H Authorization:"Bearer your-token-here" \
-d "url=https://app.matroid.com/images/logo.png"
# Classify a local image
curl -X POST "https://app.matroid.com/api/v1/detectors/detector-id/classify_image" \
-H Authorization:"Bearer your-token-here" \
-F "file=@/home/matroid/logo.png"
const configs = { localizationConfidence: 0.50, numResults: 20 };
// Classify an image from a URL
const urlImage = { url: 'https://app.matroid.com/images/logo.png' };
api.retrieveToken()
.then(token => api.classifyImage('detector-id', urlImage, configs));
// Classify a local image
const localImage = { file: '/home/matroid/logo.png' };
api.retrieveToken()
.then((token) => api.classifyImage('detector-id', localImage, configs));
# Classify an image from a URL api.classify_image(detectorId='detector-id', url='https://app.matroid.com/images/logo.png') # Classify a local image api.classify_image(detectorId='detector-id', file='/home/matroid/logo.png')
CURL
JAVASCRIPT
PYTHON
# Monitor a live video stream with a detector
curl -X POST "https://app.matroid.com/api/v1/streams/stream-id/monitor/detector-id" \
-H Authorization:"Bearer your-token-here" \
-d 'thresholds={ "cat": 0.5, "dog": 0.6 }'
// Monitor a live video stream with a detector
const configs = { thresholds: { cat: 0.5, dog: 0.6 } };
api.retrieveToken()
.then((token) => api.monitorStream('stream-id', 'detector-id', configs));
# Monitor a live video stream with a detector
api.monitor_stream(
streamId='stream-id',
detectorId='detector-id',
thresholds={ 'cat': 0.5, 'dog': 0.6 })
CURL
JAVASCRIPT
PYTHON
# Create a detector from COCO format dataset
curl -X POST "https://app.matroid.com/api/v1/detectors" \
-H Authorization:"Bearer your-token-here" \
-F "file=@/home/matroid/catdog.zip" \
-F "name=cat-dog" \
-F "detector_type=general" \
-F "labelsJSON=@/home/matroid/labels.json"
// Create detector from a COCO format dataset
api.retrieveToken()
.then((token) =>
api.createDetector(
'/home/matroid/catdog.zip',
'cat-dog',
'general',
'/home/matroid/labels.json'
)
);
# Create a detector from a COCO format dataset api.create_detector( file='/home/matroid/catdog.zip', name='cat-dog', detectorType='general', labelsJSON='/home/matroid/labels.json')
CURL
JAVASCRIPT
PYTHON
# Search for similar images in an S3 bucket
curl -X POST "https://app.matroid.com/api/v1/collection-tasks/task-id/image-query" \
-H Authorization:"Bearer your-token-here" \
-d "url=url-for-your-image" \
-d 'boundingBox={ "top": 0.1, "left": 0.1, "height": 0.8, "width": 0.8 }' \
-d "numResults=5" \
-d "shouldIndicateDuplicates=true"
// Search for similar images in an S3 bucket
const configs = {
boundingBox: { top: 0.8, left: 0.8, height: 0.1, width: 0.1 },
shouldIndicateDuplicates: true
numResults: 5,
};
api.retrieveToken()
.then((token) =>
api.queryCollectionByImage(
'task-id',
{ url: 'https://image.jpg' },
configs
)
);
# Search for similar images in an S3 bucket
api.query_collection_by_image(taskId='task-id',
url='https://image.jpg',
boundingBox={ 'top': 0.8, 'left': 0.8, 'height': 0.1, 'width': 0.1 },
numResults=5,
shouldIndicateDuplicates=True)
Capabilities
API Support
Matroid API is able to classify images, locate objects in videos, search existing collections of media, and perform real-time detections in live video streams.
A Detector is a trained computer vision model that is compatible with the Matroid platform. Matroid detectors are the center of the Matroid Universe: they can be used to search through images, videos, and streams to find objects or people of interest. Detectors can be created, trained, deleted, and retrained. Detectors can be imported into Matroid from existing models or exported to other Matroid users from the API.
Request A Demo Today
Let us show you how the power of computer vision can transform your organization