AI model development with low code! What is a “model farm”?

What is the low-code AI model development platform “Model Farm”?

Thundercomm's model farm has a graphical user interface(GUI)Low-code that enables creation of AI models with intuitive operation byA development platform.

There are various methods for AI model development, and methods of developing from scratch using deep learning frameworks such as TensorFlow and PyTorch are widely known. However,it is also true that it is very difficult to build an appropriate network after understanding the characteristics of deep learning, and we often hear that development is slow. Various issues have been raised, such as securing resources for AI engineers, expanding development costs, and lengthening development time.

With model farms,developers can manage image data, add recognition targets, and learn/train models using a GUI, using image datasets tailored to each use case, without being aware of deep learning frameworks ornetworkstructures.It is possible to proceed with the development of AI models by performing verification, simulation, etc.

End-to-end support from data set management to model training/validation

The model farm workflow is roughly divided into two phases, and data services and model services are provided as follows.

Data service: confirmation / management of additional datasets, labeling, dashboard to check the status of data preprocessing, etc.

Model service: selection of pre-trained models, setting of learning parameters, training/validation, online testing, etc.

Since many pre-trained AI models are provided on the platform, you can add any data set based on your favorite pre-trained model and proceed with training, realizing efficient model development. . learn more/After verificationQualcomm SnapdragonMounted EdgeAIIt is also possible to output a model optimized for the terminal.AIThis makes it easier to avoid problems caused by model compatibility when converting models for embedded devices, and enables a smooth transition to embedded application development.