# Machine Learning Information and Links ## Important Note The machine learning cameras use the switch on the side to toggle between training and VS connection modes. - Switch DOWN for training mode. - Switch UP for Vision System connection mode. To test a cam: Power with 5V and GND from an arduino. Look on your list of available networks. You should see a network called "WiFiCam_XXX" where XXX are the numbers on your cam (found on a sticker on the back of the camera board). If not, try removing power, switching the switch, and adding power. As always, please ask a UTF or LTF for help if you are stuck or having trouble! ## Procedure Summary Please read the detailed steps for this process [here](https://docs.google.com/document/d/e/2PACX-1vQ62vc19j6C2-0Kt96RCXP47GEOVDUF8kx1D0DLjhxCqnHA-1scFjtmLmQ8KGbFA4QTX7Vno-3M_Wva/pub). 1. Determine the categories you will classify images in to. (In the lesson, this was thumbs up/thumbs down.) 2. Didn't do the lesson/activity in class? Feel free to do the activity yourself (it takes about 20 minutes) [here](https://colab.research.google.com/drive/1oN1PaApCx8lXFLCyGs9ZnhK8Sk2WBDNH?usp=sharing). 3. Gather training images of all categories using a WiFi Cam. 4. Train a model on Google Colab using [this notebook](https://colab.research.google.com/drive/1UgAg3rmqENqiPOPe5mAHOpDjTshxs0hI?usp=sharing&authuser=1). 5. Upload the model. (See model upload below) 6. Connect the WiFi Cam to your Arduino and request predictions from your model! - If you get a "model not found" error (prediction returns 257), double check that you uploaded the model correctly. - If you're not seeing your cam on the Vision System, double check that the switch is in the UP position. ## Model Upload Click [here](https://enes100.umd.edu/tools/modeluploader) to upload a trained model. ## WIFI Cam CAD Download Cam Case with camera model, best for placing in CAD. Download Cam Case as STEP file, best for editing/customizing in CAD. Download Cam Case as STL file, best for 3D printing.

Below is an example of a possible mount to connect the camera. The hole is sized for an M4 screw.