Edge Conductor

Simplified Edge AI

One Click Model Training & Deployment

With Edge Conductor, training and deploying AI models becomes even easier. You can conveniently train AI models by selecting the desired training dataset and regularly update the models through the scheduling feature. AI models can be easily deployed in any environment where Edge App is installed. Simplify the AI model training and deployment process with Edge Conductor and experience efficient model management.

AI Monitoring

Real-Time AI Inference Monitoring

A powerful tool for monitoring real-time AI inference results and efficiently collecting inference data. This data can be used to create new training datasets, continuously improving the learning performance of AI models. With the outstanding features of Edge Conductor, you can monitor inference results in real-time, collect necessary data, and perform precise analysis and model updates. You can also selectively configure webhooks for various scenarios, such as when an Edge device disconnects or when AI model inference results return a defect. Depending on the values set by the user, real-time alerts can be received through the desired channels.

Model Optimization

Dataset Management Features for Continuous AI Model Optimization

Edge Conductor enables users to manage their data directly and optimize AI models. Users can collect real-time inference result data to create new datasets, upload data stored in their local environment, or, if needed, link data stored in the cloud to their dataset. Since the dataset feature is provided through an easy-to-use UI, even non-experts can easily manage their own data.

Building Training Datasets Through Data Re-labeling

You can efficiently re-label collected image data to improve AI model performance. By utilizing the re-labeling feature, you can train high-performance AI models with accurate and precise data. Edge Conductor simplifies image data processing and labeling tasks, providing higher-accuracy training data. This maximizes AI model training efficiency and improves data accuracy, leading to superior analysis results.

Model Retraining Process

Edge Conductor simplifies the AI model retraining process, making it easy for users to update their AI models. Users can select a new dataset and specify the desired AI model to start retraining. Once retraining is complete, the new AI model can be deployed to Edge App to keep the model up-to-date. Additionally, by leveraging the scheduling feature, retraining and deployment can be performed automatically at specified times or intervals without user intervention. The required dataset, stream, and target Edge for deployment are selected, and training and deployment occur automatically at the scheduled time. All of these processes are provided through an intuitive UI, allowing users to retrain and deploy AI models easily without complex configurations.

Architecture

Support for Various Operating Environments

Edge Conductor can be installed and used in both cloud and on-premise environments. Even in cases where data security concerns prevent storing data in the cloud, or when Edge AI must be operated within a private network, anyone can install and use Edge Conductor in their own business environment.

Workspace Structure for Managing Multiple Projects

Edge Conductor supports a workspace structure to efficiently manage multiple projects. Each workspace has Edge devices registered and managed individually, and accounts and permissions are also separated by workspace. Thanks to this structure, only users with permission to a specific workspace can access its associated Edge devices, and the data collected from Edge and used for training can also only be accessed by authorized users. This allows Edge devices and data to be securely separated and managed per project, making it suitable for protecting sensitive data. Easy permission management enables both collaboration and security. Administrators can clearly identify and control users, Edge devices, and data status for each workspace. In this way, Edge Conductor provides a safe and efficient environment for operating various projects.

Flexible and Scalable Architecture

Users can monitor the inference results of multiple Edge Apps and deploy retrained models through Edge Conductor. The training datasets needed for model retraining are generated from Edge App inference results, local data on a notebook, and data from AWS Cloud S3. Users can improve data accuracy through dataset re-labeling, enhancing model performance and reliability. The continuous inference monitoring and retrained model redeployment system ensures high-performance and reliable inference on Edge App. Defined alerts based on inference and model training/deployment status are provided on the web and also sent in real-time to users through external systems like Slack and Teams.