purelyvivid / tfjs_model Public

Notifications You must be signed in to change notification settings
Fork 0
Star 0

Generate and utilize a Pre-trained Model for Tensorflow.js

0 stars 0 forks Branches Tags Activity

Name		Name	Last commit message	Last commit date
Latest commit History 8 Commits
tfjs_mobilenet		tfjs_mobilenet
CNAME		CNAME
README.md		README.md
convert_model_from_keras_to_tfjs.py		convert_model_from_keras_to_tfjs.py
demo.png		demo.png
utilization.png		utilization.png

Repository files navigation

Generate and Utilize Model Files for Tensorflow.js

How to Generate Model Files of tfjs (inclubing Structure and Weights) ?
1. Need to install python library tensorflowjs first.
```
pip install tensorflowjs
```
1. Then convert models from Keras to Tensorflow.js (refer to here)
How to Utilize these Model Files?
1. Publish Model Files to Web and get the Link of model.json. For this case, the URL is : http://mobilenet.purelyvivid_tfjs.nctu.me/tfjs_mobilenet/model.json
2. Download https://github.com/tensorflow/tfjs-examples . In the folder of /mobilenet, open the file named index.js. Modify the const MOBILENET_MODEL_PATH = ... to the URL of your model.
1. Confirm your Environment: (could refer to here if your OS is Ubuntu)
  
  Node.js >=8.9 and <9
  
  Yarn
2. Run JS code
```
yarn
yarn watch
```
1. Then you can see the result such like:
  
  Note: the text in above image is just for modification reference, not actually true.