diff --git a/README.md b/README.md
index 3f8b6cc..d0b6dd4 100644
--- a/README.md
+++ b/README.md
@@ -27,12 +27,29 @@ Espessially for analog counter is to center the picture on the pointer.
 
 ### Read the images
 
+
+
+#### Install collectmeteranalog
+
+The [releases](https://github.com/haverland/collectmeteranalog/releases) contains downloads for Windows, Linux and MacOS. But the prefered install is via python's pip.
+
+##### Python
+
 This is mostly the easiest part, if you have installed python on your computer. If not you need to install it ( <https://www.python.org/downloads/> ).
 
 Open a terminal and type in:
 
     pip install git+https://github.com/haverland/collectmeteranalog
 
+On mac and windows the prediction is not available. It shows everytime a -1. You can manually install it by 
+
+    pip install tensorflow-macos 
+or
+
+    pip install tensorflow
+
+The application is called via console
+
     python3 -m collectmeteranalog --collect=<your-esp32name> --days=3
 
 It downloads now all images in a "data" subfolder. The image names will be hashed for your privacy.
@@ -40,6 +57,21 @@ Be patiant. It will takes a while.
 
 After it the duplicates will be automaticly removed and finally you have a folder named data/labled with the images.
 
+##### Windows, MacOS, Linux
+
+The executables are console applications. You can use it like python
+
+    Windows-collectmeteranalog.exe --collect=<your-esp32name> --days=3
+,
+
+    Linux-collectmeteranalog --collect=<your-esp32name> --days=3
+or
+
+    macOS-collectmeteranalog --collect=<your-esp32name> --days=3
+
+Windows and MacOS excecutables have no prediction, because the tflite-runtime is only available for linux and the complete tensorflow library is to big (600MB) for a single application.
+
+
 ### Label the images
 
 Now you can label the images. After reading the images it opens a window.
@@ -63,9 +95,6 @@ We want label like a human would be read the pointer.
 If it is correctly, you can click on update. If not use the slider to adjust it.
 
 The prediction on the left side can help you to identify the number. But beware the model can be only a help for you. **Don't trust the recognition!**
-On mac the prediction is not available. It shows everytime a -1. You can manually install it by 
-
-    pip install tensorflow-macos
 
 <img src="images/Labeling3.png" width="600">
 
diff --git a/setup.py b/setup.py
index 619e367..d2ab98e 100644
--- a/setup.py
+++ b/setup.py
@@ -19,7 +19,7 @@
 
 setup(
     name='collectmeteranalog',
-    version='0.2.3',
+    version='1.0.0',
     url='https://github.com/haverland/collectmeteranalog',
     license='Apache 2.0',
     author='Frank Haverland',