With the advancement of technology, artificial intelligence (AI) has become an important part of the global education system. Countries are actively promoting AI courses in primary and secondary schools to introduce students to this cutting-edge technology from an early age. Today, AI is not just the domain of scientists and engineers; it has entered classrooms as a powerful tool to enhance students' creativity and problem-solving skills.

What is Visual Machine Learning?

An important branch of AI is machine learning, which focuses on training models with data to make predictions or classifications. Machine learning platforms leverage this principle, allowing users to upload or collect data such as images, audio, and video to train models for image classification, pose recognition, sound classification, and more. This intuitive operation enables students to experience the working principles of AI firsthand and understand the underlying logic through experimentation.

KittenAI Visual Machine Learning is a new AI feature added to the Kittenblock programming platform. It aims to allow users to quickly create AI models through simple visual operations without programming experience. Unlike traditional machine learning platforms, it not only covers machine learning training but also integrates graphical programming. Once the training is complete, the results can be directly imported into the programming interface for use. Its intuitive interface and simple drag-and-drop operations make building AI models easy and fun, making it particularly suitable for primary and secondary school students to showcase their skills in AI project creation.

KittenAI Visual Machine Learning solves the challenge of implementing AI in primary and secondary schools by offering a lightweight, quick-to-experience, and fast-to-teach solution. It supports web-based access, allowing for immediate use, meeting schools' low-threshold and low-cost AI learning needs.

Principles of Machine Learning

The core of Teachable Machine is deep learning technology based on neural networks. Simply put, it "learns" the associations between data by inputting a large amount of training data (such as images, videos, or audio). The model uses these associations to predict or classify new data in the future. For students, the process of creating such deep learning models is simplified into a series of visual operations, making complex problems simpler. They first learn to use it, then further enhance their understanding and interest in the technology.

KittenAI Visual Machine Learning Process

Students can add the [Machine Learning (Visual)] plugin in Kittenblock to build and apply their own AI models. Here are the two main stages for students using this feature:

1. Build and train a machine learning model, then import it into the graphical programming interface with one click.
2. Start programming and designing graphical project programs.

Steps to Build a Visual Machine Learning Model

1. Choose the task type: Currently, there are three types available: image classification, pose classification, or audio.
2. Data collection: Students can provide training data for the model through real-time camera images or prepared images, microphones, or existing media files.
3. Train the model: With one-click training, machine learning quickly trains a usable model based on the input data.
4. Test and apply: Students can test the model's recognition effect in real-time in the preview box and then apply it to real-world scenarios.

Programming and Designing Graphical Project Programs

After machine learning training is complete and imported into the programming interface, corresponding programming blocks are available, including "current recognition result," "recognition confidence," and more. By combining these blocks with creativity, students can build various specialized AI projects, such as "waste classification," "part classification," "pest recognition," "prohibiting electric bikes from entering residential buildings," "posture monitoring," "sound classification recognition," and more.

Compared to previous machine learning that focused on the correctness of recognition results, this visual machine learning emphasizes the entire process. Through learning, application, and debugging, students can adjust learning materials based on satisfaction with the results, continuously optimizing and forming good project-based learning habits.

Advantages

1. No programming background required: The platform is designed for users with zero background, with simple and clear operations. Students can easily create AI models and quickly experience them without learning complex programming languages.
2. Intuitive visual interface: Through a graphical interface, students can see the model's training process and results in real-time, enhancing the intuitiveness and fun of learning.
3. Quick and convenient, ready to use: Using the web version, there are no cumbersome downloads. Just open the webpage to start learning and teaching.
4. Localized data training: Students can collect data based on their living scenarios and train models, enhancing the personalization and practicality of learning.
5. Integration with graphical programming: Through online hardware interaction, it can achieve software and hardware integration, broadening creative bottlenecks.

Applications

Image Classification: Students can train models to recognize different objects by taking photos of various objects or uploading different types of images. For example, they can train AI to recognize everyday items like pencils, erasers, and books. When they present these objects, the machine learning model can accurately identify the categories of the items.
Pose Classification: Using a camera, students can capture different poses and train the model. For instance, they can set different human poses, and when they perform these poses, the system can recognize and classify them.
Audio Classification: Students can record different sounds, such as clapping, speaking, or tapping on a table. After training the model, the system can recognize these different actions or sound categories through audio.

Through these practical examples, students can not only understand the application scope of AI but also apply these skills to solve real-life problems.

Getting Started

1. Open the web version (currently, new features are only adapted for the web version) at https://kblock.kittenbot.cc/.

2. Add the extension and select Machine Learning (Visual).

3. Choose from three types of machine learning.

4. For example, in image classification, input data for "scissors," "rock," and "paper," then train and use the model.

5. Use simple programming logic.

6. Integrate hardware control.

7. This completes the process from machine learning training to programming application.

Extensions

Due to space limitations, not all possibilities can be explored. By combining image, audio, and pose classification with real-life scenarios, students can create many interesting projects, such as:

Waste Classification Assistant: Students can train a model to recognize recyclable and non-recyclable waste by taking photos of various types of waste, helping family members with waste sorting.
Home Security Detector: Students can train an AI model to detect unusual sounds in the home by recording different sounds, such as knocking or glass breaking. When the system recognizes these sounds, it can trigger an alarm.
Sports Training Detection: Students can create an AI model to recognize different human poses, helping to determine whether specific sports poses meet standards.
Overall, this new visual machine learning tool is very suitable for primary and secondary school students to learn and engage with AI. It simplifies complex AI operations and uses practical examples to help students understand how AI works and apply it to everyday life.

By using visual machine learning, the complexity of AI courses is greatly reduced. Teachers can use this tool to let students explore the world of AI hands-on. This not only enhances students' interest in AI but also helps them develop stronger technological literacy for their future studies and careers.