REST API for production inference and open source models for download. Free to start.
Production-ready REST API for all KINEVA models. One call is enough. Free tier included.
URL or Base64. Any image format.
CNN, PCNN, or VLLM. Choose model or automatic.
Bounding boxes, classes, confidence. Structured and immediately usable.
# Detect objects in an image curl -X POST https://api.kineva.eu/v1/detect \ -H "Authorization: Bearer YOUR_API_KEY" \ -H "Content-Type: application/json" \ -d '{ "model": "vehicle-detection-v3", "image": "https://example.com/street.jpg", "confidence_threshold": 0.5 }'
// Response { "model": "vehicle-detection-v3", "detections": [ { "class": "car", "confidence": 0.97, "bbox": [120, 340, 480, 560] } ], "inference_time_ms": 42 }
Download select KINEVA models for free. Run on your own infrastructure, fine-tune with your data, deploy at the edge.
Accurate head and person counting in crowded scenes. Works from any camera angle.
Detects license plates in real-time. GDPR-compliant anonymisation mode included.
Detects and classifies passenger cars and commercial trucks for traffic monitoring.
General-purpose 80-class baseline for rapid prototyping and transfer learning.
Road sign detection with GPS-tagged geolocation for infrastructure monitoring.
Head-based face anonymisation. GDPR-compliant, on-device processing.
PCNN model with WiFi probe and GPS fusion. Deployed across 60+ cities.
# Install from Hugging Face pip install huggingface_hub # Download a model from huggingface_hub import hf_hub_download model_path = hf_hub_download( repo_id="rebotnix/kineva-head-detection-v3", filename="model.onnx" ) # Run inference import onnxruntime as ort session = ort.InferenceSession(model_path) results = session.run(None, {"input": image_tensor})
API access or commercial model license. We respond within 24 hours.