Dr. Fabian Bause from Beckhoff discusses their new AutoML tool for vision tasks in the industrial sector. They explore the evolution of company offerings, integration with Beckhoff tools, target audience analysis, AutoML capabilities, model accuracy, and customer feedback.
AutoML simplifies AI model creation for domain experts without deep ML knowledge.
Beckhoff's AutoML tool supports vision applications and offers platform-aware model creation.
Deep dives
AutoML as an Enabler for Domain Experts
AutoML is presented as an enabler for domain experts, providing them with the ability to create AI models without the in-depth knowledge of machine learning concepts. By automating the data scientist's tasks, AutoML simplifies the process for domain experts, such as machine building engineers, to generate AI models. The emphasis is on allowing users to create AI models without delving into the complexities of machine learning, making it accessible to a broader audience.
Expansion of Back of's AI Capabilities
Back of Automation has evolved its portfolio to encompass more sophisticated AI capabilities, particularly focusing on vision applications. The enhancement includes the ability to process image data and deploy deep learning models on GPUs in addition to CPUs, highlighting the company's transition into a comprehensive vision company. This expansion aims to facilitate the integration of AI technology in an industrial environment, covering both hardware and software aspects.
Openness and User-Friendly Interface of the AutoML Tool
The AutoML tool by Back of is designed to be platform-aware, allowing users to select specific hardware configurations and latency limits for model execution. By providing a user-friendly interface, the tool simplifies the AI model creation process, ensuring ease of use for individuals unfamiliar with AI technologies. Moreover, the tool offers transparency post-model generation, enabling users to inspect statistical metrics and visualizations to understand the model's behavior, fostering trust and acceptance among users.