Neurala has developed Inspector for AITRIOS to work with the Lucid Labs Triton Smart camera, featuring Sony's IMX501 image sensor with AI processing capabilities.

Inspector for AITRIOS is similar to Neurala's standard Inspector application. The following documentation will be helpful in understanding how to use Inspector for AITRIOS.

• Using Inspector
• Capturing Datasets
• Importing Brains
• Running Inspections

There are some key differences in Inspector for AITRIOS, so please read this page carefully to understand how to use the application.

📷 Connect the Triton Smart camera

Before connecting the Triton Smart camera to Inspector, note the following requirements:

1. Make sure the Triton Smart camera has the most recent firmware installed.
2. The computer and camera must be on the same network.
3. The camera MTU needs to be set above 9000 (above computer interface), otherwise the camera initialization will fail and the Inspection won't start. In Windows, this setting is called "Jumbo Packets."

To set the camera as an Input in Inspector, follow the instructions in Using Inspector. On the Inputs screen, Inspector should recognize the connected Triton Smart camera. Select it and save it as an Input.

⤵️ Importing & loading Brains

In Neurala's standard Inspector application, images are captured from a connected camera and inspections are run using a Brain on the PC.

In Inspector for AITRIOS, the Brain is loaded directly onto the camera and most of the inspection process is completed there. Inspector receives the output from the camera and displays the inspection results for the user.

The process to load Brains onto the Triton Smart camera and import them into Inspector involves a few steps:

1. Export the trained Brain from Brain Builder for AITRIOS. See Brain Builder for AITRIOS.
2. Load the Brain onto the Triton Smart camera using the tools provided by Lucid Labs:
   1. Use the LucidConverterTool to convert and package the exported Brain. (This tool can be obtained from Lucid Labs.)
   2. Use the Lucid Labs Arena software to load the converted Brain onto the Triton Smart camera. This is done using the firmware update page in ArenaView.
3. Import the Brain into Inspector. On the Brains page, click the + button to import the Brain. From the menu select Choose to select the Brain file that you exported from Brain Builder. (Do not click Import from Brain Builder, as it will not work.) Use the file menu to locate the zip file that was exported from Brain Builder.

⚠️

A few key notes:

Be sure to import the same Brain into Inspector that you used with the Lucid applications to convert and load onto the camera.

When a Brain has been converted with the LucidConverterTool, it will only work on the camera that was specified during the conversion process. If you want to run that same Brain on a different Triton Smart camera, it will need to be converted and packaged again.

4. Configure the Brain output settings according to the instructions on Importing Brains.

🔍 Running inspections

On the Inspect page, you should see the Triton Smart camera as an available input.

1. Use the + button to add a Brain to the camera. Use the drop-down to select the same Brain that you already loaded onto the camera using the Arena software.
2. After the Brain has been added to the input, click Start Inspecting 1 Input. This will begin the inspection process.

⏲️

The Triton Smart camera has to "initialize" the Brain the first time you run the inspection. This process will take 30-60 seconds in most cases.

Inspections will then run, and the images and results can be viewed by clicking on the Live View link.

🚧

If Inspector shows an error when trying to access live view, it may mean another operations is still controlling the camera. If this occurs, power cycle the camera and reboot the PC to resolve it.

To stop inspections, click the Stop Inspecting button.

🤔 Camera connection troubleshooting

There may be scenarios in which Inspector is not able to appropriately connect to or recognize the Triton Smart camera. The following steps may be helpful to resolve this problem.

Linux

Windows