https://github.com/marcogdepinto/panoramamodelserving

Using Java and deeplearning4j to serve a Keras model.
https://github.com/marcogdepinto/panoramamodelserving

deep-learning deep-neural-networks deeplearning deeplearning4j java keras python

Last synced: 17 days ago
JSON representation

Using Java and deeplearning4j to serve a Keras model.

• Host: GitHub
• URL: https://github.com/marcogdepinto/panoramamodelserving
• Owner: marcogdepinto
• Created: 2019-08-04T13:33:38.000Z (over 5 years ago)
• Default Branch: master
• Last Pushed: 2022-10-05T19:31:25.000Z (about 2 years ago)
• Last Synced: 2024-10-30T06:05:52.939Z (2 months ago)
• Topics: deep-learning, deep-neural-networks, deeplearning, deeplearning4j, java, keras, python
• Language: Java
• Homepage:
• Size: 563 KB
• Stars: 1
• Watchers: 3
• Forks: 0
• Open Issues: 3
• Metadata Files:
  • Readme: README.md

Awesome Lists containing this project

README

        
# Keras model deploying with Java

Starting from another project I made to classify images using Keras (https://github.com/marcogdepinto/ScenarioClassifier), I had the idea to work on a Deploy class in Java. 

Deploy means placing the model in production: this class can be, for example, included in a Spring Boot Application and consumed via API calls.

The library used is [DeepLearning4j](https://deeplearning4j.org/).

# Model

Please download the model from [this link](https://drive.google.com/open?id=1jPDQcqQeh7r-_yQgn9jJzc8yVXEOgkvk) (it exceeds 100 megabytes so it can't stay on Github) and place it in the resource folder.

# Output

![city](https://github.com/marcogdepinto/Java-KerasDLModelServing/blob/master/city2.jpg)

Passing the above picture as input, the class will return the following information

```

[city, desert, mountain, nature, sea, universe]

[[    1.0000,         0,         0,         0,         0,         0]]

[1.0, 0.0, 0.0, 0.0, 0.0, 0.0]

city

```

**Output explanation**

- The first line is an array with the labels.

- The second line includes an array with the predicted percentage per class.

- The third line is given as output to match the labels and the array of predictions.

- The forth line is the prediction in human-readable language.