About the sample AI models locations:
The Triton Smart requires an AI model to be loaded before it can produce an input tensor. The input tensor characteristics depend on the type of AI model that is loaded:
- Classifier AI model uses a 256 x 256 px input tensor
- Detector AI model uses a 320 x 320 px input tensor
- Anomaly Detection uses: 512 x 512 px for Fine / Ultrafine, 256 x 256 px for Normal / Coarse
The Triton Smart Support package ZIP file contains pre-trained models that adhere to the above specifications. You need to upload the specific pre-trained model that corresponds to your planned model type.
- Sample Classifier AI model is located in \Networks\BrainBuilder\CL_cat_dog\BB_output\
- Sample Detector AI model is located in \Networks\BrainBuilder\OD_cat_dog\BB_output\
- Sample Anomaly Detection model is located in \Networks\BrainBuilder\AH_hazelnuts)\BB_output
You will need the following files:
| File: | Location: |
|---|---|
network_info.txt (Used to create fpk_info.dat) | TritonSmart_Support_Package_v#.#.#\Networks\BrainBuilder\***\BB_output |
BB_network.fpk | TritonSmart_Support_Package_v#.#.#\Networks\BrainBuilder\***\BB_output |
labels.txt | TritonSmart_Support_Package_v#.#.#\Networks\BrainBuilder\***\BB_output |
loader.fpk | TritonSmart_Support_Package_v#.#.#\imx501_files |
firmware.fpk | TritonSmart_Support_Package_v#.#.#\imx501_files |
*** = Choose one of the three folder to use.