sv-v5 / mobilenet-img Goto Github PK

Use a pretrained model to extend an image classifier and then predict examples using TFJS.

Drag-and-drop a folder named training (included in this repo in data/) with examples organized by labels in separate folders.

├── landscape
│   ├── Beautiful-winter-thick-snow-trees-dusk_2560x1600.jpg
│   ├── dnipro_house.jpg
│   ├── silhouette_mountains.jpg
│   ├── witcher_3_wild_hunt_landscape_mountains.jpg
│   └── ws_Banff_Park_Beautiful_Winter_1920x1080.jpg
└── not-landscape
    ├── dark-bkgrnd.png
    ├── fridge_sv.jpg
    ├── lambo_night.png
    └── thinking-face.png

Drag-and-drop a folder named validation (included in this repo in data/) with examples organized by labels in separate folders.

├── landscape
│   └── lxqt_landscape.jpg
└── not-landscape
    └── brain-comp-chip.png

The MobileNet classifier app is hosted here: https://sv-v5.github.io/mobilenet-img

npm start to compile and test locally

npm run build to build and optimize for production

• TensorFlow MobileNet models
• Kevin Scott for the app and UI ml-classifier-ui
• rutvikokate for an example of the app