One-Click Deploy

One-click deploy is Workbench's single-button wrapper for the whole chain AIHub model → board filesystem → usable in the Vision tab. The whole flow takes about 30–90 seconds (depending on model size and network).

What it does

After you press "Deploy to board", the IDE automatically does 4 things:

  1. Downloads the .xmm.zip model package from AIHub
  2. Pushes it over SSH to the Pico-G1 board at /mnt/models/xmm/
  3. Extracts and verifies its integrity on the board
  4. The model appears automatically in the Vision tab's shimeta-xmm panel, ready to run with one click

Where to find it

The "Deploy to board" button at the bottom of each model card, in both the Model Market tab and the User Models tab:

The "Deploy to board" button on a model card — Model Market tab

The "Deploy to board" button on a model card — User Models tab

Full flow

User picks a model in the Models tab
       │
       ▼  clicks "Deploy to board"
The IDE pushes it automatically over SSH:
  scp <model>.xmm.zip → shimetad@192.168.49.10:/mnt/models/xmm/
       │
       ▼
The board extracts it to /mnt/models/xmm/<model>/
       │
       ▼
Vision tab → the model appears automatically in the shimeta-xmm panel
       │
       ▼  clicks "Run"
The shimeta-xmm service starts
       │
       ▼
The "Object view" box in the upper right shows the live camera feed

Step by step

1. Pick a model

Browse the cards in the ch03 Model Market or the User Models sub-tab.

2. Click "Deploy to board"

The button sits at the bottom of each model card.

3. Wait for deployment to finish

The IDE automatically:

StepAction
1Pushes the .xmm.zip over SSH to the board at /mnt/models/xmm/
2Extracts it on the board
3Verifies the model's integrity

After a successful deployment the button changes to "Deployed".

4. Switch to the Vision tab

Click the top "Vision" tab → the model appears automatically in the shimeta-xmm panel on the right.

5. Set parameters + run

In the xmm panel on the right:

ParameterDefaultDescription
conf0.25Confidence threshold
iou0.45NMS IOU threshold
ClassesAllChoose which classes to detect

Click "Run" → the camera feed appears in the "Object view" box in the upper right.

Offline deployment

If the board has no internet access, you can manually scp a locally downloaded .xmm.zip through the Workbench Terminal tab:

# In the IDE Terminal tab
scp /path/to/<model>.xmm.zip shimetad@192.168.49.10:/mnt/models/xmm/

The deployment history UI is not yet complete in v1.0.0; to check deployment status, SSH into the board and inspect the /mnt/models/xmm/ directory.

Troubleshooting

SymptomFix
Deploy button unresponsiveBoard not connected; go back to ch02
Deployment stuck at "pushing model"SSH unreachable; check the IP / network
Deployed but nothing shows in the Vision tabThe model's .xmm file is corrupted; deploy again
Deployment fails with "not enough disk space"SSH into the board and clean up old models in /mnt/models/xmm/

Next