Build focused AI models
Train separate AI models for objects, faces, hand gestures, pose gestures, drawings, text, and number prediction.
General Image AI
Use the full image for simple image classification.
Face AI
Use a face focus area so the model mainly learns the face.
Hand Gesture AI
Train gestures like open hand, fist, OK, thumbs up.
Pose Gesture AI
Train body poses using a full-body focus area.
Object AI
Train objects using a selectable focus box.
Shape / Alphabet AI
Train digits, alphabets, symbols, and shapes.
Text AI
Classify messages, topics, emotions, and intents.
Number AI
Predict values from numeric data.
Focused Vision AI
General Image AI learns the full image. Face, hand, pose, and object sections use a focus area/crop so unwanted background has less effect on training.
Train using the full camera image
Best for simple object/image classification when background is controlled.
Train using face-focused samples
Place the face inside the focus box. The saved examples are cropped to the face area.
Train hand gestures
Place only the hand gesture inside the focus box. Face and room background are mostly ignored.
Train body pose gestures
Keep the body inside the focus area. Useful for hands up, T-pose, sitting, standing, and exercise poses.
Train object recognition
Put the object inside the focus box before adding examples, so background affects the model less.
Train alphabets, digits, shapes, or symbols
Students draw examples on canvas. The model learns from their own drawings.
Classes
Drawing Canvas
Train & Test
Train text classification
Useful for sentiment, topic, spam/not spam, and chatbot intent activities.
Classes & Examples
Train & Test
Train number prediction
Enter input/output data and train a simple regression model.
Training Data
| Input X | Output Y |
|---|
Predict
Save projects and models
Download the complete project or individual model files for MadBot AI Studio.
Save Complete Project
Save all sections, classes, examples, and trained model data into one JSON file.
Load Project
Upload a previously downloaded MadBot ML Lab project JSON file.
Important Notes
- Camera works only on HTTPS domains such as train.madbot.in or on localhost.
- Face, Hand, Pose, and Object AI use focus/crop areas to reduce background confusion.
- Add a negative class such as No Hand, No Face, No Object, or Empty.