Update to Versioni 4.1.0

* Upgrade to Tensorflow 2.0
* Usage of Pillow in CNN-Image Processing

Author: jomjol

Changelog.md

@@ -1,4 +1,8 @@
 ## Changelog
+##### 3.0.2 (2019-11-07)
+* Correct error for disable Logging
+##### 3.0.1 (2019-11-03)
+* Update to Pillow 6.2.0
 ##### 3.0.0 (2019-10-06)
 * Impementation of optional consistency check of readout value (not negative, maximum rate)
 ##### 2.3.0 (2019-10-05)

README.md

@@ -4,12 +4,10 @@ This repository is the sum of different projects to read out an analog water met
 The result is a HTTP-server, that takes an image as input, processes it and gives as an output the water meter number, including the subdigits.
 
 ## Changelog - lastest version
-##### 3.0.2 (2019-11-07)
-* Correct error for disable Logging
-##### 3.0.1 (2019-11-03)
-* Update to Pillow 6.2.0
-##### 3.0.0 (2019-10-06)
-* Impementation of optional consistency check of readout value (not negative, maximum rate)
+##### 4.1.0 (2019-11-25)
+* Upgrade to Tensorflow 2.0
+* Changed image processing within CNN to Pillow (instead of OpenCV)
+
 ### [Full Changelog](Changelog.md)
 
 
@@ -87,6 +85,7 @@ Example with parameter `full`:
 
 The page `roi.html` return the image including the ROIs visible. This is usefull to check for correct setting:
 <img src="./images/roi_masked.jpg" width="400">
+Since version 4.1.0 also a middle cross and a circle for supporting the alignment is implemented.
 
 * http://server-ip:3000/setPreValue.html?value=401.57
 
@@ -96,9 +95,9 @@ The page `setPreValue.html` stores the number given in the parameter `value` to
 | --------- | ------- | ------- |
 | value | valid setting of water meter readout | `value=401.57` |
 
+* http://server-ip:3000/version.html
 
-
-
+The page `version.html` returns a version number.
 
 
 

code/Dockerfile

@@ -3,15 +3,14 @@ FROM python:3.7
 WORKDIR /
 
 COPY ./tensorflow ./tensorflow
-RUN pip install --no-cache-dir ./tensorflow/tensorflow-1.14.0-cp37-cp37m-linux_x86_64.whl
+RUN pip install --no-cache-dir ./tensorflow/tensorflow-2.0.0-cp37-cp37m-linux_x86_64.whl
 RUN rm -r ./tensorflow
 
 COPY requirements.txt ./
 RUN pip install --no-cache-dir -r requirements.txt
 
 COPY . ./
-
-CMD chmod +x ./restart.sh
+RUN rm -r ./tensorflow
 
 EXPOSE 3000
 

code/config/config.ini

@@ -2,7 +2,7 @@
 TimeoutLoadImage=30
 #IP durch die IP des ESP32 ersetzen!
 URLImageSource=http://192.168.178.22/capture_with_flashlight
-LogImageLocation=./log/source_image
+#LogImageLocation=./log/source_image
 # Falls nur schlechte / fehlerhafte Bilder gelockt werden sollen, naechste Zeile Kommentar entfernen
 #LogOnlyFalsePictures=True
 
@@ -40,14 +40,14 @@ pos_y=386
 
 [Digital_Digit]
 names=ziffer1, ziffer2, ziffer3, ziffer4, ziffer5
-Modelfile=./config/neuralnets/Train_CNN_Digital-Readout_Version2.h5
-LogImageLocation=./log/digital_digit
+Modelfile=./config/neuralnets/Train_CNN_Digital-Readout_Version_4.1.0.h5
+#LogImageLocation=./log/digital_digit
 #LogNames=zeiger3, zeiger4
 
 [Analog_Counter]
 names=zeiger1, zeiger2, zeiger3, zeiger4
-Modelfile=./config/neuralnets/CNN_Analog-Readout_Version-2.1.1.h5
-LogImageLocation=./log/analog_counter
+Modelfile=./config/neuralnets/CNN_Analog-Readout_Version-4.0.1.h5
+#LogImageLocation=./log/analog_counter
 #LogNames=zeiger3, zeiger4
 
 [Analog_Counter.zeiger1]

code/config/neuralnets/CNN_Analog-Readout_Version-2.1.1.h5

@@ -2,12 +2,12 @@
 TimeoutLoadImage=30
 #IP durch die IP des ESP32 ersetzen!
 URLImageSource=http://192.168.178.22/capture_with_flashlight
-LogImageLocation=./log/source_image
+#LogImageLocation=./log/source_image
 # Falls nur schlechte / fehlerhafte Bilder gelockt werden sollen, naechste Zeile Kommentar entfernen
 #LogOnlyFalsePictures=True
 
 [ConsistencyCheck]
-Enabled=False
+Enabled=True
 AllowNegativeRates=False
 #Maximum Change of new to old value (+ or -)
 MaxRateValue=0.1
@@ -40,14 +40,14 @@ pos_y=386
 
 [Digital_Digit]
 names=ziffer1, ziffer2, ziffer3, ziffer4, ziffer5
-Modelfile=./config/neuralnets/Train_CNN_Digital-Readout_Version2.h5
-LogImageLocation=./log/digital_digit
+Modelfile=./config/neuralnets/Train_CNN_Digital-Readout_Version_4.1.0.h5
+#LogImageLocation=./log/digital_digit
 #LogNames=zeiger3, zeiger4
 
 [Analog_Counter]
 names=zeiger1, zeiger2, zeiger3, zeiger4
-Modelfile=./config/neuralnets/CNN_Analog-Readout_Version-2.1.1.h5
-LogImageLocation=./log/analog_counter
+Modelfile=./config/neuralnets/CNN_Analog-Readout_Version-4.0.1.h5
+#LogImageLocation=./log/analog_counter
 #LogNames=zeiger3, zeiger4
 
 [Analog_Counter.zeiger1]