We Keep Coding

How to pass input data to an existing tensorflow 2.x model in Java?

I'm doing my first steps with tensorflow. After having created a simple model for MNIST data in Python, I now want to import this model into Java and use it for classification. However, I don't manage to pass the input data to the model.
Here is the Python code for model creation:
from tensorflow.keras.datasets import mnist
from tensorflow.keras.utils import to_categorical.
(train_images, train_labels), (test_images, test_labels) = mnist.load_data()
train_images = train_images.reshape((60000, 28, 28, 1))
train_images = train_images.astype('float32')
train_images /= 255
test_images = test_images.reshape((10000, 28, 28, 1))
test_images = test_images.astype('float32')
test_images /= 255
train_labels = to_categorical(train_labels)
test_labels = to_categorical(test_labels)
NrTrainimages = train_images.shape[0]
NrTestimages = test_images.shape[0]
import os
import numpy as np
from tensorflow.keras.callbacks import TensorBoard
from tensorflow.keras.models import Sequential
from tensorflow.keras.layers import Dense, Dropout, Flatten
from tensorflow.keras.layers import Conv2D, MaxPooling2D
from tensorflow.keras import backend as K
# Network architecture
model = Sequential()
mnist_inputshape = train_images.shape[1:4]
# Convolutional block 1
model.add(Conv2D(32, kernel_size=(5,5),
activation = 'relu',
input_shape=mnist_inputshape,
name = 'Input_Layer'))
model.add(MaxPooling2D(pool_size=(2,2)))
# Convolutional block 2
model.add(Conv2D(64, kernel_size=(5,5),activation= 'relu'))
model.add(MaxPooling2D(pool_size=(2,2)))
model.add(Dropout(0.5))
# Prediction block
model.add(Flatten())
model.add(Dense(128, activation='relu', name='features'))
model.add(Dropout(0.5))
model.add(Dense(64, activation='relu'))
model.add(Dense(10, activation='softmax', name = 'Output_Layer'))
model.compile(loss='categorical_crossentropy',
optimizer='Adam',
metrics=['accuracy'])
LOGDIR = "logs"
my_tensorboard = TensorBoard(log_dir = LOGDIR,
histogram_freq=0,
write_graph=True,
write_images=True)
my_batch_size = 128
my_num_classes = 10
my_epochs = 5
history = model.fit(train_images, train_labels,
batch_size=my_batch_size,
callbacks=[my_tensorboard],
epochs=my_epochs,
use_multiprocessing=False,
verbose=1,
validation_data=(test_images, test_labels))
score = model.evaluate(test_images, test_labels)
modeldir = 'models'
model.save(modeldir, save_format = 'tf')
For Java, I am trying to adapt the App.java code published here.
I am struggling with replacing this snippet:
Tensor result = s.runner()
.feed("input_tensor", inputTensor)
.feed("dropout/keep_prob", keep_prob)
.fetch("output_tensor")
.run().get(0);
While in this code, a particular input tensor is used to pass the data, in my model, there are only layers and no individual named tensors. Thus, the following doesn't work:
Tensor<?> result = s.runner()
.feed("Input_Layer/kernel", inputTensor)
.fetch("Output_Layer/kernel")
.run().get(0);
How do I pass the data to and get the output from my model in Java?

With the newest version of TensorFlow Java, you don't need to search for yourself the name of the input/output tensors from the model signature or from the graph. You can simply call the following:
try (SavedModelBundle model = SavedModelBundle.load("./model", "serve");
Tensor<TFloat32> image = TFloat32.tensorOf(...); // There a many ways to pass you image bytes here
Tensor<TFloat32> result = model.call(image).expect(TFloat32.DTYPE)) {
System.out.println("Result is " + result.data().getFloat());
}
}
TensorFlow Java will automatically take care of mapping your input/output tensors to the right nodes.

I finally managed to find a solution. To get all the tensor names in the graph, I used the following code:
for (Iterator it = smb.graph().operations(); it.hasNext();) {
Operation op = (Operation) it.next();
System.out.println("Operation name: " + op.name());
}
From this, I figured out that the following works:
SavedModelBundle smb = SavedModelBundle.load("./model", "serve");
Session s = smb.session();
Tensor<Float> inputTensor = Tensor.<Float>create(imagesArray, Float.class);
Tensor<Float> result = s.runner()
.feed("serving_default_Input_Layer_input", inputTensor)
.fetch("StatefulPartitionedCall")
.run().get(0).expect(Float.class);

Feature Layer is getting null bounds with featurecollection derived from RasterToVectorProcess

Null Pinter Exception simply means that it is because of value is getting null. But In case of using APis such as GeoTiff it becomes annoying to find out the error in usage.
My code is as follows:
System.out.println("vectorization starts");
GridCoverage2D srcCoverage = new GeoTiffReader(new File("E:/output/ll_processed.TIFF")).read(new GeneralParameterValue[]{policy, gridsize, useJaiRead});
SimpleFeatureCollection fc = RasterToVectorProcess.process(srcCoverage, 3, cov.getEnvelope(), Collections.singletonList(0.0d), true, null);
System.out.println("process ends");
System.out.println("vectorization ends");
//MapContext map = new DefaultMapContext();
//map.setTitle("raster to vector conversion");
Style style = SLD.createPolygonStyle(Color.BLUE, Color.CYAN, 1.0f);
//map.addLayer(fc, style);
//map.getLayerBounds();
//JMapFrame.showMap(map);
MapContent mapContent= new MapContent();
mapContent.setTitle("Illegal Mining");
Layer layer = new FeatureLayer(fc, style,"VectorLayer");
//int boundary = 10;
// ReferencedEnvelope env2 = new ReferencedEnvelope(srcCoverage.getEnvelope().getMinimum(0) - boundary, srcCoverage.getEnvelope().getMaximum(0) + boundary,
//srcCoverage.getEnvelope().getMinimum(1) - boundary, srcCoverage.getEnvelope().getMaximum(1) + boundary, srcCoverage.getCoordinateReferenceSystem());
//mapContent.getViewport().setBounds(fc.getBounds());
Line 199 : if(layer.getBounds()!=null) // here the error is coming also tried with if(layer != null && layer.getBounds()!=null)
{
mapContent.addLayer(layer);
}else{
System.out.println("Layer bounds are null");
}
mapContent.getViewport().setCoordinateReferenceSystem(
DefaultGeographicCRS.WGS84);
Error
at org.geotools.map.FeatureLayer.getBounds(FeatureLayer.java:199)
I am trying to convert Tiff to Vector image and Then I want to store it on disk.

Most likely, your featureSource is null, since the geotools FeatureLayer class method getBounds() uses the featureSource to retrieve the bounds, but in the constructor FeatureLayer doesn't check whether the featureSource is null.
The featureSource is the first argument to the constructor of FeatureLayer. In your case, that's variable SimpleFeatureCollection fc.
Most likely, the method process.process returned null so fc is null.
Minor Change in RasterToVectorProcess.java
//features.add(builder.buildFeature(null));
//System.out.println("ignored");
//add
System.out.println("adding...");
SimpleFeature feature = builder.buildFeature(null);
((Collection<SimpleFeature>) features).add(feature);

Java - Won't read out File content

I have a .txt file with, for example, this content:
variable1="hello";
variable2="bye";
testing3="parameter";
whatisthis4="hello";
var5="exampletext";
example=3;
wellthen=8;
---
It read in the file, line by line, fine until I added a way of saving the data.
This whole code plus another reader (with other variable names of course) is wrapped in a try-catch statement.
String path_playlist = new File("").getAbsolutePath();
String fileName_playlist = path_playlist
+ "/src/dancefusion/game/playlist.txt";
FileReader fr_playlist = new FileReader(fileName_playlist);
BufferedReader br_playlist = new BufferedReader(fr_playlist);
int track_counter = track_sum*9;
String trackinfos[] = new String[track_counter];
while(track_counter < 0)
{
System.out.println("linecount="+track_counter);
trackinfos[track_counter] = br_playlist.readLine();
System.out.println(trackinfos[track_counter]);
track_counter--;
}
System.out.println(Arrays.toString(trackinfos));
In this example track_sum equals 1.
The while loop should read in the file one line at a time but only reads null's:
[null, null, null, null, null, null, null, null, null]
Update 1:
The while-condition was set up the wrong way... thanks!
The corrected version:
while(track_counter < 0)
However, now it gives me an exception with an "ArrayOutOfBounds: 9".
Any guesses?
Final Update:
As mentioned by #GiorgiMoniava, I just needed to reduce track_counter by one before starting to read in as in Java arrays begin with 0, thanks!
int track_counter = track_sum*8;
String trackinfos[] = new String[track_counter];
track_counter--;
while(track_counter >= 0)
{
System.out.println("linecount="+track_counter);
trackinfos[track_counter] = br_playlist.readLine();
System.out.println(trackinfos[track_counter]);
track_counter--;
}
Maybe one of you can figure out what I did wrong...
Of course I can deliver more information/code if needed!
Thanks in advance!

This looks weird
while(track_counter < 0)
Are you sure loop is ever entered in? It is my guess (from your output) that track_counter is 9.
About your array out of bounds exception: if you create array of size N you can only access it using indexes: [0, N-1]

try this
while(track_counter > 0)
{
System.out.println("linecount="+track_counter);
track_counter--;
trackinfos[track_counter] = br_playlist.readLine();
System.out.println(trackinfos[track_counter]);
}

calling a tell and told prolog predicate from java does not work

I have the below code and when I run it through Prolog it creates a new file with a new rule. For example, I run
create_rec(check_symptoms(Symptoms,noOk))
it creates a new file with name rule_new and content all predicates with the old rules
and the new rule predicate with content is
rule(r15,_G3583,_G3601):-check_symptoms(_G3593,noOk)
The problem is that when I call the predicate create_rec from Java it does not work:
rules([r1,r2,r3,r4,r5,r6,r7,r8,r9,r10,r11,r12,r13,r14]).
% other predicates
create_rec(Body) :-
clause(rules(R_ids), B1),
last(R_ids,LastRule_id),
atom_codes(LastRule_id, Rule_codes),
Rule_codes = [H|T],
atom_codes(AtomNum,T),
atom_number(AtomNum,Num),
NewNum is Num+1,
atom_codes(NewNum,NewNumCodes),
atom_codes(NewRule_id,[114|NewNumCodes]),
append(R_ids, [NewRule_id], New_R_ids),
retract(rules(R_ids)),
asserta((rules(New_R_ids))),
asserta((rule(NewRule_id,A,B) :- Body)),save_rule.
save_rule :-
tell('rule_new.pl'),
write(':- dynamic rule/3, rules/1.'),nl,
write(':- [\'kb_anemia_V5b.pl\'].'),nl,
write(':- encoding(utf8).'),nl,
write(':- style_check(-singleton).'),
clause(rules(R_ids),B),nl,
write((rules(R_ids) :- B)),
write('.'),
get_rule_data(R_ids),
told.
get_rule_data([]).
get_rule_data([Rule_id|Rest_Rule_Id]) :-
clause(rule(Rule_id,A,B),Body1),
% fix(B,[],B2),
write(rule(Rule_id,A,B2):-Body1),write('.'), nl,
get_rule_data(Rest_Rule_Id).
% other predicates
The code in Java is:
Term consult_arg[] = {
new Atom(Diagnosis.class.getResource("anemia_diagnosis").getPath())};
Query consult_query = new Query( // to kanei query gia na ginei
"consult", //to consult
consult_arg);
boolean consulted = consult_query.hasSolution();
if (!consulted) {
System.err.println("Consult failed");
System.exit(1);
}
bodycr = body_txt1.getText();
String t9 = "create_rec(" + bodycr + " )." + "\n";
System.out.println("FUNCTION IS " + t9);
Query q9 = new Query(t9);
diagnosis = q9.oneSolution().toString();
System.out.println(diagnosis);
JOptionPane.showMessageDialog(null, "KB is created ");
It takes the body from the textarea and call the predicate. The pop up message is displayed , no bugs were found but the file is not created.
I don't know what is wrong. Can anyone help me?

JT400 ProgramCall's run() method is not returning any result

I am new to JT400. I am trying to invoke a test program in AS400 through JT400. Here is my code
public class TestRpg {
public static void main(String[] args){
try{
AS400 sys=new AS400("mydomain","username","password");
String number="asdf <= Return value from Java Input";
String lnsts="";
String amount="";
String lnofcd="";
AS400Text txt80 = new AS400Text(80);
AS400Text txt50 = new AS400Text(50);
ProgramParameter[] parmList = new ProgramParameter[4];
parmList[0] = new ProgramParameter( txt80.toBytes(number),80);
parmList[1] = new ProgramParameter( txt50.toBytes(lnsts),50);
parmList[2] = new ProgramParameter( txt80.toBytes(amount),80);
parmList[3] = new ProgramParameter( txt50.toBytes(lnofcd),50);
ProgramCall pgm = new ProgramCall(sys,"/QSYS.LIB/mylib.LIB/testrpg.PGM",parmList);
if (pgm.run()!=true) {
System.out.println("executed");
}else{
System.out.println("Output Data 0: " + (String)txt80.toObject( parmList[0].getOutputData() ) );
System.out.println("Output Data 1: " + (String)txt50.toObject( parmList[1].getOutputData() ) );
System.out.println("Output Data 2: " + (String)txt80.toObject( parmList[2].getOutputData() ) );
System.out.println("Output Data 3: " + (String)txt50.toObject( parmList[3].getOutputData() ) );
sys.disconnectService(AS400.COMMAND);
}
AS400Message[] messageList = pgm.getMessageList();
System.out.println(messageList.length);
for (int i=0; i < messageList.length; i++)
{
System.out.print ( messageList[i].getID() );
System.out.print ( ": " );
System.out.println( messageList[i].getText() );
}
sys.disconnectService(AS400.COMMAND);
}catch(Exception e) {
System.out.println(e.toString());
}
}
}
I had debug the code it's not giving any response after executing
pgm.run(). It is not even showing any exception. Programme is just holding at pgm.run() and not returning any thing.
As per the comments I got, I want to include the scenario I am trying to work on. In AS400 when we execute the testrpg.pgm program, it displays a screen with four input fields and some function keys to perform operations. My intention is to invoke f2 function key of that program from JT400. Is the approach I am following is the right way? Please suggest me

All program calls happen in batch so your program is most likely in MSGW on the server. Find it with wrkactjob and investigate the message it is waiting for, and give the appropriate action.
This is typically due to incorrectly formed parameters.

This is a common misunderstanding, so just for clarification for other readers:
Calling a Cobol/RPG program from Java is batch, just the same as calling the Cobol/RPG program from a Cobol/RPG/CL.
How to begin: Create a program which you can call from CL:
... declare and fill MYFIELD1, MYFIELD2 ...
CALL PGM(MYPGM) PARM(&MYFIELD1 &MYFIELD2)
...
If this works, it will also work from Java using jt400, if you:
call the right AS400 using correct credentials
call the right program in the right library
use the right number and lentgh of parameters
In case of crash as described (waiting forever), DSPMSG QSYSOPR will show an open message, like "MCH0801 = wrong number of parameters". D=Dump will create a spoolfile where you see which incoming parameters are filled with which content, or you see "undefined".