Image Classification Application on GENI Network via Socket Programming.

0. Demo

Open your browser and visit

http://198.248.248.136:8080/Image-App/

If prompted for password, username: admin, password: 123.

Upload your image (size < 1MB, tested with .jpg).

Live Demo:

https://youtu.be/rSX9DdryEFA

1. Setup Inference Server

Step 1 Reserve Geni Slice

Create a geni slice using the rspec in GitHub repo: fin_request_rspec.xml

Log into the node called Infer.

Step 2 Setup Requirements

git clone https://github.com/TTTheo/Image-Recognition-Application.git

cd Image-Recognition-Application/NNServer/

bash setup_server.sh

source ~/.bashrc

Test whether java is install correctly:

java -version
>> penjdk version "13.0.1" 2019-10-15
>> OpenJDK Runtime Environment (build 13.0.1+9)
>> OpenJDK 64-Bit Server VM (build 13.0.1+9, mixed mode, sharing)

javac -version
>> javac 13.0.1

Test whether conda and python is installed correctly:

conda --version
>> conda 4.8.0

python --version
>> Python 3.7.4

Step 3 Run Server

Move to the inference server directory:

cd Image-Recognition-Application/NNServer/

Make temporary folders for input and output queues:

mkdir input
mkdir output

Compile java code:

javac InferenceServer.java

Run the socket server to listen to incomming connections

nohup java InferenceServer 8888 &

Press enter to get back the terminal prompt.

Run the neural network model with ResNet50:

nohup python RunInfer.py 50 &

Press enter to get back the terminal prompt. If it is your first time running it, you should wait for 1-2 minutes while the script is downloading the trained model.

Now the the inference server should be up running and listening to incomming connections and ready to run the nerual network model to recognize images. The public IP address of the server could be found using ifconfig command (which is required at the web server to connect to the inference server).

When you are finished you should kill both java and python because they should be starting and running at the same time.

killall python
killall java

2. Setup Web Server

Step 1 Enter Web Server Node

Log into the node called Web

Step 2 Setup Requirements

Install Java:

sudo su
wget https://raw.githubusercontent.com/TTTheo/Image-Recognition-Application/master/Java-steup.sh
bash Java-steup.sh
echo 'export JAVA_HOME=~/java/jdk-13.0.1' >> ~/.bashrc 
echo 'export PATH=$JAVA_HOME/bin:$PATH' >> ~/.bashrc 
source ~/.bashrc

Test whether java is install correctly:

java -version
>> penjdk version "13.0.1" 2019-10-15
>> OpenJDK Runtime Environment (build 13.0.1+9)
>> OpenJDK 64-Bit Server VM (build 13.0.1+9, mixed mode, sharing)

javac -version
>> javac 13.0.1

Install TomCat

wget https://raw.githubusercontent.com/TTTheo/Image-Recognition-Application/master/Tomcat-setup.sh
bash Tomcat-setup.sh

bash tomcat/apache-tomcat-9.0.29/bin/startup.sh

sudo mv tomcat/manager.xml tomcat/apache-tomcat-9.0.29/conf/Catalina/localhost/
sudo mv tomcat/tomcat-users.xml tomcat/apache-tomcat-9.0.29/conf/

exit

Step 3 Run Server

Download GitHub repo which contains teh WAR file we need to set up the web server:

git clone https://github.com/TTTheo/Image-Recognition-Application.git

Get the routable-IP of web server node and visit the webpage:

http://Routable-IP:8080/manager/html

Login with username: admin and password: 123

Deploy -> Deploy directory or WAR file located on server WAR or Directory path:

/users/peilun/Image-Recognition-Application/FileUpload/target/Image-App.war

Click Deloy

Finally visit:

Then visit http://Routable-IP:8080/Image-App