Fix setup.py to not include the direct link to tflite_runtime wheels (so that we can publish on PyPi). Add instructions for installing tflite_runtime, and fix the Raspberry Pi install script to reflect those new instructions.

mbeissinger · mbeissinger · commit 656261b0e434 · 2021-08-06T19:33:57.000-07:00
diff --git a/README.md b/README.md
@@ -1,7 +1,7 @@
 # Lobe Python API
 Code to run exported Lobe models in Python using the TensorFlow, TensorFlow Lite, or ONNX options.
 
-Works with Python 3.6, 3.7, or 3.8, untested for other versions.
+Works with Python 3.6, 3.7, and 3.8 untested for other versions.
 
 ## Install
 ### Backend options with pip
@@ -13,8 +13,9 @@ pip install lobe[all]
 # For TensorFlow only
 pip install lobe[tf]
 
-# For TensorFlow Lite only
-pip install lobe[tflite]
+# For TensorFlow Lite only -- this requires two steps for the runtime and for lobe (note for Raspberry Pi see our setup script in scripts/lobe-rpi-install.sh)
+pip install --index-url https://google-coral.github.io/py-repo/ tflite_runtime 
+pip install lobe
 
 # For ONNX only
 pip install lobe[onnx]
@@ -72,7 +73,7 @@ pip install lobe[all]
 ```python
 from lobe import ImageModel
 
-model = ImageModel.load('path/to/exported/model')
+model = ImageModel.load('path/to/exported/model/folder')
 
 # OPTION 1: Predict from an image file
 result = model.predict_from_file('path/to/file.jpg')
diff --git a/examples/raspberry-pi/README.md b/examples/raspberry-pi/README.md
@@ -24,12 +24,13 @@ It is assumed you are starting with a Pi in a remote headless configuration alre
 2. Power up your Pi and log in to the desktop view.
 3. On the Pi, open a terminal window and create a directory where you want to store your Lobe model and code. For example, in the /home/pi folder:
     ```
-        mkdir Lobe
+    mkdir Lobe
     ```
-4. Navigate into your Lobe directory, and download the setuptools and the lobe-python package:
+4. Navigate into your Lobe directory, and install the lobe-python package:
     ```
-        pip3 install setuptools
-        pip3 install git+https://github.com/lobe/lobe-python
+    cd Lobe
+    wget https://raw.githubusercontent.com/lobe/lobe-python/master/scripts/lobe-rpi-install.sh
+    sudo ./lobe-rpi-install.sh
     ```
 
 ## Getting the Lobe TF Model on your Pi
@@ -52,8 +53,4 @@ It is assumed you are starting with a Pi in a remote headless configuration alre
 ### Troubleshooting
 1. Check that the Pi power light is bright red. A dim red light indicates insufficient power.
 2. Be sure you're installing and running your TF code using Python 3.
-3. If the TensorFlow module is not recognized, try re-installing:
-    ''' 
-        pip3 install tensorflow == 1.13.1
-    '''
-4. If you're using multiple components, be sure they are all connected to the same common ground.
+3. If you're using multiple components, be sure they are all connected to the same common ground.
diff --git a/scripts/lobe-rpi-install.sh b/scripts/lobe-rpi-install.sh
@@ -1,6 +1,8 @@
 #!/bin/bash
-#
-# Install Python3
+# Add TensorFlow Lite package repo (https://www.tensorflow.org/lite/guide/python#install_tensorflow_lite_for_python)
+echo "deb https://packages.cloud.google.com/apt coral-edgetpu-stable main" | sudo tee /etc/apt/sources.list.d/coral-edgetpu.list
+curl https://packages.cloud.google.com/apt/doc/apt-key.gpg | sudo apt-key add -
+# Install Python3 and deps
 sudo apt -y update
 sudo apt install -y \
     python3-dev \
@@ -11,5 +13,7 @@ sudo apt install -y \
     libjpeg62-turbo
 sudo apt-get install -y git
 sudo pip3 install setuptools
-# Install lobe-python with TensorFlow Lite backend
-sudo pip3 install lobe[tflite]
+# Install TensorFlow Lite Runtime
+sudo apt install -y python3-tflite-runtime
+# Install lobe-python which can use the tflite backend
+sudo pip3 install lobe
diff --git a/setup.py b/setup.py
@@ -78,9 +78,9 @@
     package_dir={"": "src"},
     install_requires=requirements,
     extras_require={
-        'all': [tf_req, onnx_req, tflite_req],
+        'all': [tf_req, onnx_req],
         'tf': [tf_req],
         'onnx': [onnx_req],
-        'tflite': [tflite_req],
+        #'tflite': [tflite_req],
     }
 )
diff --git a/src/lobe/backends/backend_tflite.py b/src/lobe/backends/backend_tflite.py
@@ -12,7 +12,7 @@
 
     except ImportError:
         raise ImportError(
-            "ERROR: This is a TensorFlow Lite model and requires TensorFlow Lite interpreter to be installed on this device. Please install lobe-python with lobe[tflite] or lobe[all] options. If that doesn't work, please go to https://www.tensorflow.org/lite/guide/python and download the appropriate version for you device."
+            "ERROR: This is a TensorFlow Lite model and requires TensorFlow Lite interpreter to be installed on this device. Please go to https://www.tensorflow.org/lite/guide/python#install_tensorflow_lite_for_python for installation instructions for you device."
         )
 
 

Original file line number	Diff line number	Diff line change
`@@ -78,9 +78,9 @@`
`78`	`78`	`package_dir={"": "src"},`
`79`	`79`	`install_requires=requirements,`
`80`	`80`	`extras_require={`
`81`		`- 'all': [tf_req, onnx_req, tflite_req],`
	`81`	`+ 'all': [tf_req, onnx_req],`
`82`	`82`	`'tf': [tf_req],`
`83`	`83`	`'onnx': [onnx_req],`
`84`		`- 'tflite': [tflite_req],`
	`84`	`+ #'tflite': [tflite_req],`
`85`	`85`	`}`
`86`	`86`	`)`
Original file line number	Diff line number	Diff line change
`@@ -12,7 +12,7 @@`
`12`	`12`
`13`	`13`	`except ImportError:`
`14`	`14`	`raise ImportError(`
`15`		`- "ERROR: This is a TensorFlow Lite model and requires TensorFlow Lite interpreter to be installed on this device. Please install lobe-python with lobe[tflite] or lobe[all] options. If that doesn't work, please go to https://www.tensorflow.org/lite/guide/python and download the appropriate version for you device."`
	`15`	`+ "ERROR: This is a TensorFlow Lite model and requires TensorFlow Lite interpreter to be installed on this device. Please go to https://www.tensorflow.org/lite/guide/python#install_tensorflow_lite_for_python for installation instructions for you device."`
`16`	`16`	`)`
`17`	`17`
`18`	`18`