Hey, I am the programmer for team #17658 Wildcat Robotics. I am having trouble getting the HuskyLens to focus on the artifacts and recognize them. What is the best way to get the HuskyLens to recognize them? Would uploading models of them via a microSD card be better or something else?
Thank you!
Would uploading models of them via a microSD card
I don’t know if you’re interpreting the term “model” correctly. That’s not a “3D Model”, that’s a “Neural Network Model”. You can only upload models trained to recognize ARTIFACTS if you’ve previously saved a working model that recognizes ARTIFACTS.
This video (https://youtu.be/E140gPLPz4A?t=1333) shows an example of training the HuskyLens to recognize objects. I would recommend teaching the HuskyLens how to recognize both green and purple ARTIFACTS as you likely want them to be considered different objects.
-Danny
Thanks for the advice! I just wasn’t sure what model meant and the huskylens has had some difficulty with recognizing the artifacts. I was trying to get it to recognize the separate colors. I’ll definitely look into the video you sent. Thanks again!
For color-specific recognition, training huskylens in object mode for each artifact color is the easiest workaround. Put the item in a well-lit place, hold it still, and “teach” it from different angles. Lighting and backdrop contrast matter more than most people think—a busy background will confuse it. Huskylens don’t generalize well, so don’t use a preloaded microSD model unless you have one trained for your artifacts. You may test in real life and adjust distance and angles once you recognize each hue. object mode + numerous teaches generally fixes the issue without changing SD card models.