Data Spree offers a no‑code AI platform that lets manufacturing and logistics teams build, train, and deploy computer‑vision models without programming. The web‑based solution handles data collection, automated labeling, one‑click model training, and export to ONNX for edge or cloud inference on hardware such as Intel OpenVINO, NVIDIA Jetson, and Movidius.
Funding
Funding not disclosed
Founders
Product
Problem
Industrial manufacturers and logistics operators often rely on manual visual inspection and custom‑built computer‑vision systems that require extensive programming, specialized hardware, and long development cycles, leading to high costs, low scalability, and delayed defect detection.
Solution
Data Spree provides a no‑code AI Platform that streamlines the entire computer‑vision workflow—from data collection and annotation to model training, validation, and deployment. Users can connect cameras or sensors, automatically generate labeled datasets, and train state‑of‑the‑art deep neural networks with a few clicks. Trained models are exported in ONNX format and run on the Data Spree Inference engine, which supports edge and cloud hardware, including Intel OpenVINO, NVIDIA Jetson, and Movidius. The platform includes real‑time monitoring, statistical data management, and modular pipelines, enabling rapid adaptation to new products or process changes without programming.
Target Audience
Primary customers are engineering teams in manufacturing, logistics, and quality‑inspection departments that need fast, scalable computer‑vision solutions without deep AI expertise.
Features
- Web‑based interface for image upload, automated labeling (object tracking, annotation status) and quality‑assured ground‑truth creation
- Integrated data management with real‑time analytics, statistics, and historical tracking of defects and performance metrics
- One‑click training of classification and object‑detection models with automated hyper‑parameter optimization and data augmentation
- Real‑time training monitoring with loss, precision, recall, and confusion matrix visualizations
- Export of models in ONNX format and support for multiple runtimes (Python, C/C++, Java) via Data Spree Inference
- Vendor‑agnostic camera and sensor integration (2D, Time‑of‑Flight, stereo, laser scanners) through REST API and plugin architecture
- Multi‑model and multi‑sensor pipelines with customizable Python plugins and built‑in heatmap visualizations
- Deployment on edge devices (Intel CPUs/GPUs, NVIDIA Jetson, Movidius) and cloud servers with Intel OpenVINO acceleration