Deep Learning DS empowers engineers to work on image processing projects faster and more efficiently by using the Data Spree AI workflow to accomplish key computer vision tasks in just hours, not months. By delivering automated data labeling features, data analytics, automated parameter optimization and state-of-the-art AI algorithms everyone can now exploit the full potential of neural networks and independently develop comprehensive solutions for new and varied computer vision challenges for every enterprise in every industry.
1. Connect cameras, sensors and other sources with Deep Learning DS to acquire data.
2. Data acquisition: So-called ground truth data form the basis for the successful training of AI models.
3. AI models are trained and evaluated with a just a few mouse clicks.
4. Deploy AI for real-time execution using Inference DS.
5. The fully trained AI can be quickly integrated into existing and newly designed systems.
6. High quality data is generated and analyzed during AI execution (inference).
AI solutions are based on data. In Deep Learning DS the creation of image data sets and valid ground truth data are easy and fast. Any kind of image data, wether industry camera data, point clouds or even image data from your phone, can be uploaded, managed and labeled. Deep Learning DS offers an efficient and time-saving image labeling process. Object tracking and automated data labeling functions help to deal with large-scale image data sets. A quality assurance process is integrated to ensure high data quality for good training results.
You can easily upload data to Deep Learning DS via the intuitive user interface or integrate it into your systems via the REST-API. Automatic data collection is available via Inference DS.
All data at a glance. Deep Learning DS makes it easy to manage every kind and size of image data. Statistical evaluations about data sets, annotations, classes and parameters in real time are part of the extensive data management feature set. Only a transparent overview of the data enables valid AI development.
Deep Learning DS always provides state-of-the-art AI models for different computer vision tasks, like classification or object detection. Every model is optimized and fits different requirements according to accuracy, latency and capacity.
Without any programming or machine learning skills required AI trainings can be started, tracked and evaluated by the press of a button. Data augmentation and parameter optimization features can be used manually or fully automated.
During the model training the progress and quality of the training can be monitored live. Losses, precision and recall are shown in charts during the training. During the training the model evaluation is done via automatically generated charts showing the level of accuracy, precision, recall and confusion. Different training results and models can be generated and compared in parallel. A reliable evaluation is the basis for a reliable problem solution.
The trained AI model can be exported quickly and easily. Deep Learning DS supports the open ONNX standard for easy integration of the trained models.
With Inference DS, you can run AI models on different hardware.