Is there a way to make the HUSKYLENS recognize an object that isn't one of the 20 built in objects?
Samuel.Cuevas 2024-10-28 20:00:11 2148 Views3 Replies Im trying to get the Huskylens to recognize a different object from the 20 that come built in with the camera, is there a way to teach the Huskylens a new object?
Isn't it possible to store learned objects in non-volatile internal memory ?
claan719 Yes, the HUSKYLENS can recognize objects beyond the 20 built-in categories using its learning mode. The steps to achieve this depend on what type of object recognition you want (e.g., color, shape, or a specific custom object). Here's how you can approach it:
1. Using the Learning Mode
HUSKYLENS includes a "Learning" mode that allows you to teach it to recognize custom objects. Here's how to use it:
Switch to the appropriate algorithm mode (e.g., "Object Recognition" or "Custom Object Recognition").Point the HUSKYLENS camera at the object you want it to learn.Press the "Learn" button (either on the device itself or through your connected controller/software interface).Move the object or adjust its angle to ensure the HUSKYLENS captures various perspectives.The HUSKYLENS will store the learned object in its memory and recognize it in subsequent views.2. Optimizing Recognition
To improve the recognition of custom objects:
Ensure good lighting conditions and minimal background noise.Provide multiple angles and distances during the learning process.If the object has distinct patterns or colors, use the "Color Recognition" or "Shape Recognition" modes for enhanced accuracy.3. Saving and Reloading Custom Models
The HUSKYLENS can save learned objects in its memory, but it will lose them if powered off unless you save the data externally (e.g., using an Arduino, Raspberry Pi, or similar microcontroller).
4. Advanced Options
If the built-in recognition capabilities are not sufficient, consider integrating HUSKYLENS with a microcontroller or computer for custom-trained AI models. You can:
Use HUSKYLENS for capturing object data.Process the data externally with platforms like TensorFlow Lite or OpenCV.Reintegrate the custom recognition model with your HUSKYLENS-enabled system.Let me know if you'd like more specific guidance based on your project setup!
Waqas.Tech 
