Get started with Local Studio
Use Brain Builder for AITRIOS to train and export models for simple, small-scale use cases locally on your PC or server. Use Inspector for AITRIOS to visualize inference results and output predictions using industrial protocols.
Brain Builder for AITRIOS quick start
Build AI models with no coding, using minimal data, and easy setup on your PC or server.
*Only available for Console V1 (coming soon for Console V2).
Install Brain Builder for AITRIOS
Download, install and activate your license
Have one of the following supported edge devices ready:
Raspberry Pi AI Camera
Sony AIH-IVRW2
Rayprus CSV26
Triton Smart
Download the Brain Builder for AITRIOS installation package and CodeMeter User Runtime to activate and manage your license.
Video tutorial
Brain Builder for AITRIOS installation and license activation
Related docs
Create a dataset
Capture images for training your model
For best results, try to capture your own training images in your specific production environment. To do so, please check the table below and use the tool compatible with your edge device:
Raspberry Pi AI Camera | AIH-IVRW2 | CSV26 | Triton Smart |
|---|---|---|---|
Raspberry Pi AI Camera | AIH-IVRW2 Console/Local Console | CSV26 Console/Local Console | Triton Smart |
Dataset requirements for each AI model type:
Classifier
• Only one object/product in each image
• Min. resolution: 256x256px
• 50+ images per class
Detector
• Variations of the objects and multiple objects in each image
• Min. resolution: 320x320px
• 50+ images and 150+ tagged objects per class
Anomaly Hi-Fi
• Only one object/product in each image
• Controlled image conditions.
• Min. resolution: 256x256px (coarse and regular defects) and 512x512px (fine and ultra-fine defects)
• No. of images: 50+ images for "Normal" and 5-10 images of anomalies
Note: Anomaly Hi-Fi is not currently available for Local Edition.
Related docs
Train AI models
Select one of the AI model templates
Select the type of AI Model you want build from Classifier, Detector, and Anomaly Recognizer/Hi-Fi Models:
Classifier
Classifies images into predefined classes, providing labels ranked by confidence
Detector
Identifies and locates multiple objects within an image.
Anomaly Hi-Fi
Identifies anomalies and it can also localize areas of possible anomalies with a heatmap (Hi-Fi).
Upload your training dataset, annotate, train, evaluate and fine-tune your model in Brain Builder for AITRIOS. When you are satisfied with the evaluation results, export the optimized classifier, detector, or anomaly model package.
Video tutorials
Build a Classifier AI Model
Build an Object Detection AI Model
Build an Anomaly Detection AI Model
Related docs
Deploy to AI camera
Deploy and run your AI models
Deploy your exported AI model directly to your edge device that is ready for inference.
Use the compatible tool for your device to deploy and visualize the inference results of your model:
Raspberry Pi AI Camera | AIH-IVRW2 | CSV26 | Triton Smart |
|---|---|---|---|
Raspberry Pi AI Camera | AIH-IVRW2 Console/Local Console | CSV26 Console/Local Console | Triton Smart |
Related docs
Inspector for AITRIOS quick start
Manage the Triton Smart cameras that run AI models developed in Brain Builder for AITRIOS, visualize results, and output predictions.
Set up your device
Get your device ready, connect and update it
Have the following edge device ready:
Triton Smart
Download and install ArenaView MP from LUCID Vision Labs.
Connect the camera to your PC using ArenaView MP and make sure you update the camera firmware to the latest version. Additionally, check that you have the latest version for the IMX500 chip firmware. In case you don't, download it from here.
Related docs
Install Inspector for AITRIOS
Download, install and connect to Brain Builder for AITRIOS
Download and install Inspector for AITRIOS on the same PC as ArenaView MP.
Connect Inspector for AITRIOS to Brain Builder for AITRIOS and choose the type of connection - on the local network or cloud.
You will need also need to:
• Define your inputs by connecting your camera to Inspector for AITRIOS
• Define the output protocol - Inspector for AITRIOS supports Modbus TCP, Ethernet/IP, or Neurala's C++ plugin
• Choose data folder location
• Set environment variable for inspection
Related docs
Create a dataset
Capture images for training your model
Prepare the objects you wish to capture. You can use the feature for capturing images directly from the connected cameras in Inspector for AITRIOS. For best results, try to capture your own training images in your specific production environment.
Dataset requirements for each AI model type:
Classifier
• Only one object/product in each image
• Min. resolution: 256x256px
• 50+ images per class
Detector
• Variations of the objects and multiple objects in each image
• Min. resolution: 320x320px
• 50+ images and 150+ tagged objects per class
Anomaly Hi-Fi
• Only one object/product in each image
• Controlled image conditions.
• Min. resolution: 256x256px (coarse and regular defects) and 512x512px (fine and ultra-fine defects)
• No. of images: 50+ images for "Normal" and 5-10 images of anomalies
Related docs
Train AI models
Select one of the AI model templates
Launch Brain Builder for AITRIOS, start a project, and select the type of AI Model you want build from Classifier, Detector, and Anomaly Recognizer/Hi-Fi Models:
Classifier
Classifies images into predefined classes, providing labels ranked by confidence.
Detector
Identifies and locates multiple objects within an image.
Anomaly Hi-Fi
Identifies anomalies and it can also localize areas of possible anomalies with a heatmap (Hi-Fi).
Upload your training dataset, annotate, train, evaluate and fine-tune your model in Brain Builder for AITRIOS. When you are satisfied with the evaluation results, export the optimized classifier, detector, or anomaly model package.
Video tutorials
Build a Classifier AI Model
Build an Object Detection AI Model
Create an Anomaly Recognizer/Hi-Fi AI Model
Related docs
Deploy to Triton Smart Camera
Deploy and run your AI models
After exporting your AI model from Brain Builder for AITRIOS, make sure you have the ZIP file with the model on the PC with Inspector for AITRIOS and ArenaView MP installed.
To visualize the results, you will need to deploy your AI model to both the Triton Smart camera and to Inspector for AITRIOS.
You will be able to see the inference result in the Inspect screen from Inspector for AITRIOS.
