AI Photo

Artificial intelligence (AI) can classify various things.
This application finds your favorite photo with AI.
First, let AI learn what you like and what you do not like.
Next, AI will find you the photo you recommend.
If you apply this, you can find your favorite clothes, favorite ramen, favorite sweets, good product defectives etc.


Main function
- Deep learning
- Image learning and classification by artificial intelligence
- Image processing by filter
- Display AI learning result
- Input of users favorite image

Usage:
(1) connection and setting of image processing
1. Add a filter.
  You can add a filter by clicking the "+" icon.
2. Expand the connector area and connect the filter.
  Clicking on the connector area (the gap between the filter and the filter) expands the connector area.
  Dragging the connector end highlights the connectable connector end and connects when dragging is completed.
3. Configure the filter.
  When you click on the filter, the setting screen is displayed and you can make detailed settings.

(2) Learning and classification
1. Learning and classification filter Click to start learning artificial intelligence and display the result.
2. Confirm the result.
  Output result: Banana closer to 1.0, more apple as closer to -1.0 as apple.
  Input layer: Displays the input value used for learning.
  Intermediate layer, output layer: Indicates the learned result (weight, offset) of each phase.
  Heat map: Indicates reaction to input.

Filter type:
- Cut (clip) filter: Cut and reduce the image.
- Frequency analysis (FFT) filter: Convert images to frequency images.
- Brightness adjustment (normalization) filter: Adjust the brightness of the image.
- Blur (blur) filter: Blur the image.
- Edge extraction (sharpening) filter: Adjust brightness.
- Single color conversion (grayscale) filter: Convert RGB image to Gary image.
- Color (color) filter: Convert RGB image to Gary image around specific color.
- Artificial intelligence (neural network) filter: learn / classify Gary image with neural network.
  Activation functions: TANH, ReLU, Sigmoid, Liner, Hard TANH, Hard Shrink, Soft Shrink, Soft Plus
- Learning classification: Perform learning / classification with artificial intelligence filter.