HUSKYLENS 2 Self-Learning Classifier Function Description

1.Introduction to Self-Learning Classifier

This feature allows capturing multi-angle images of any object, learning, and recognizing any custom object.

2.Self-Learning Classification Guide

In this section, we will learn how to use HUSKYLENS 2 to classify specified objects through self-learning.

2.1 Selecting the Self-Learning Classification Function

Power on HUSKYLENS 2, wait for startup completion, and locate the "Self-Learning Classification" function.

2.2 Learning Objects and Observing Effects

Learning Objects: Align HUSKYLENS 2 with the object to be learned, ensuring it fits within the orange frame, then press Button-A at the top-right corner of HUSKYLENS 2 to learn the object.

Recognizing Objects: If a learned object is detected, the screen will frame it with a colored box and display "name: IDx Confidence" at the top.
For example, "Object:ID1 95%". Here, "name" defaults to "Object" (for naming, see "Parameter Settings"); "ID1" refers to the first learned object; "91%" indicates a confidence level of 91%. Confidence level refers to a model's "certainty" in its prediction result.
"Object:ID1 91%" means under this function, the model considers this object has a 91% probability of being the object with ID 1 that was learned.

Multi-Angle Object Learning: When learning an object within the detection frame, long-press Button-A without releasing, and adjust the viewing angle of HUSKYLENS 2 to continuously learn the object from multiple angles. After learning is completed, aim at the identifiable object and press the learning button again to continue adding to the learned objects.

During multi-angle learning, the learning progress is displayed. As shown in the figure below, "Learning 38:ID1" appears on the detection frame, where "38" means that 38 images have been learned for the object with ID1.

3.Self-Learning Classifier Parameter Settings

The factory default parameters of HUSKYLENS 2 meet basic functional requirements. For more refined functionality, you can manually adjust various parameters. The following parameters are all based on the "Self-Learning Classifier" function, so first ensure you have entered the "Self-Learning Classifier" function, as shown in the figure. To select the parameter to modify, you can slide left or right across the parameter labels below the screen.

3.1 Forgetting IDs

To forget all previously learned objects: Step 1, tap "Forget ID" on the screen; Step 2, the "Forget All IDs" pop-up window will appear, then tap "Yes" to confirm.

At this step, aim HUSKYLENS 2 at the object to be "forgotten". If HUSKYLENS 2 does not frame this object, it indicates "forgetting" is completed.

3.2 Threshold

The threshold controls the sensitivity of object recognition. A lower threshold makes the "determination of whether an object is a pre-trained object" more lenient (it is more likely to mistake unlearned objects for pre-trained objects, but rarely misses recognizing pre-trained objects). A higher threshold, on the other hand, makes the standard stricter (it is less likely to make false judgments, but may fail to recognize pre-trained objects).

Setting steps: Tap "Threshold", and a parameter adjustment slider will appear above it. Slide left to decrease the value; slide right to increase it. The effect is shown in the figure.

3.3 Set Name

This parameter allows you to set names for learned objects, which can be in both Chinese and English.

Setting steps: Tap "Set Name", scroll up and down the number in the top-left corner to select the ID for which to set a name; tap the on-screen keyboard to input the name, as shown in the left figure; after completing, tap the √button in the bottom-right corner to save; if saved successfully, a green checkmark will appear in the top-right corner.

3.4 Display Name

This parameter controls whether the name of a learned object is displayed when recognized, with the default setting being "display".

Setting steps: Tap "Display Name" to activate the toggle above it, which will show as blue (indicating the "on" state). When an object is recognized, its name will be displayed, as shown in the left figure. Tap the toggle again to switch it to the "off" state, which will appear as white. In this state, the name of the recognized object will not be displayed, as shown in the right figure.

3.5 Restore Default

This parameter restores all settings to their default states and forgets the learned IDs and names, while not clearing the exported models (see below for information on exporting models).

Setup Steps: Click "Restore Default", and after the "Restore Default Configuration" pop-up appears, click "Yes".

3.6 Exporting the Model

This parameter allows you to save and export the currently set parameters, learned IDs, and configured names to the local memory of HUSKYLENS 2. It is suitable for scenarios such as migrating parameters to another HUSKYLENS 2. This operation does not require a TF card.

Export Steps: Tap "Export Model". When the "Save Configuration To" pop-up appears, scroll the number up and down to select the model number to save (up to 5 models can be saved), then tap the "Confirm" button in the lower-left corner of the pop-up to save. After confirmation, the export will start automatically, as shown in the figure.

To view the exported model: After the "Exporting" pop-up window disappears, you can view the exported model file on your computer.

First, connect the HUSKYLENS 2 to your computer's USB port.

Next, via your computer, you can access the memory of HUSKYLENS 2 using the path shown in the following figure. You can view two model-related files with extensions .json and .bin. The numbers before the extensions are the "model numbers" selected when saving the configuration. Both files can be copied and pasted to other locations.

3.7 Importing Models

This feature allows you to import a model from HUSKYLENS 2 (referred to as "HUSKYLENS A" below) into another HUSKYLENS 2 (referred to as "HUSKYLENS B" below). This way, HUSKYLENS B can copy the objects learned and adjusted parameters from HUSKYLENS A without needing to reconfigure parameters or retrain.

Import Steps:
Step 1: Connect HUSKYLENS A to your computer, then copy the exported file to your desktop.

Step 2: Connect HUSKYLENS B to your computer, and paste the file from the previous step into the designated folder of HUSKYLENS B, as shown in the figure. (If the "self-learning-classification" folder is not found, please first perform Step 3, which will automatically create this folder when the model is imported, then return to Step 2.)

Step 3: First, confirm that you have entered the "Self-Learning Classifier" function, then click "Import Model". After the "Load Configuration" pop-up window appears, slide the slider up and down to select the model number, which should match the model number saved in the previous step.

For example, if the model file pasted into HUSKYLENS 2 is config1.json, select number 1. Finally, click "Yes" to import, and wait for the pop-up window "Loading" to disappear to confirm the import is complete.

Next, verify whether the parameters of HUSKYLENS B and HUSKYLENS A, as well as the learned objects, are consistent. As shown in the left figure below is the recognition status of HUSKYLENS B before model importation, and the right figure is the recognition status after model importation.