💎 Stone vs Jade Classification
Upload a photo and let a small CNN guess: ordinary stone or jade? (the model runs in your browser)
or try a held-out test image:
Loading model…
Your image
Prediction: –
About this model
- Task: binary image classification, Stone (0) vs Jade (1)
- Input: RGB image resized to 150 × 150 pixels, scaled to 0–1
- Model: 4 × (Conv2D + MaxPooling) → Flatten → Dense(512) → Dense(1, sigmoid), 3.45 M parameters, trained with TensorFlow/Keras and exported to TensorFlow.js (float16 weights)
- Training: Adam optimizer, binary cross-entropy, early stopping on validation accuracy — 98.4 % validation accuracy
- Course project: Machine Learning 2026/1, KMITL — Pitchakorn Boonchote (67011267), Kanchayut Paojumroon (67011101)