How the Animal Face Test Works

The feature uses image classification to estimate visual similarity categories from a face image.

1. CNN-based feature extraction

Images are processed as pixel data. A convolutional neural network extracts hierarchical visual features such as contours and patterns.

2. Model inference

At prediction time, the model returns probabilities by class. The highest probability is presented as the primary result.

3. On-device processing and privacy

Inference is executed in the browser using TensorFlow.js-compatible runtime paths when available. Uploaded images are intended for local analysis, not persistent server storage.

4. Interpretation limits

Results can vary by lighting, angle, expression, and training-data bias. Use outputs as entertainment-oriented suggestions, not diagnostic conclusions.