How juancardona05/clasificador-conejos-aves works
This is a small browser-based classifier, not a backend application. The documented workflow starts with a user-selected image or camera frames, prepares pixels in the browser, runs a TensorFlow.js model, and displays the predicted class and confidence percentages. All implementation responsibilities map to index.html because the README identifies it as containing the interface and prediction logic; the model architecture and weights are separate files. Relationships below are based on README documentation only; no source excerpts were available.
Main entry points: index.html TensorFlow.js runtime · model.json Model architecture · group1-shard1of1.bin Trained weights
Components 11
Image input
- Image selection
index.html - Camera capture
index.html - Canvas preview
index.html
Prediction
- Pixel preprocessing
index.html - TensorFlow.js runtime
index.html - Model architecture
model.json - Trained weights
group1-shard1of1.bin
Results
- Class interpretation
index.html - Prediction display
index.html
Other components
- User
- Device camera
Connections 11
- User → Image selection: selects or drops
README.md - User → Camera capture: starts camera
README.md - Device camera → Camera capture: provides frames
README.md - Image selection → Canvas preview: previews image
README.md - Camera capture → Canvas preview: previews frames
README.md - Canvas preview → Pixel preprocessing: supplies pixels
README.md - Pixel preprocessing → TensorFlow.js runtime: provides tensor
README.md - TensorFlow.js runtime → Model architecture: loads model
README.md - Model architecture → Trained weights: references weights
README.md - TensorFlow.js runtime → Class interpretation: returns output
README.md - Class interpretation → Prediction display: updates result
README.md