[Object Detection by Tensorflow] 物件辨識

Here is just do some note for how to setup and install the object detection by Tensorflow on Raspberry Pi 3B+.

The original article here

Ok, let’s go my practise.

1. Update the Raspberry Pi

sudo apt update
sudo apt upgrade
sudo apt autoremove

2. Install Tensorflow, OpenCV and related Python modules

Refer to here
You should install this tool as well.

sudo apt install protobuf-compiler

3. Setup the Tensorflow models

mkdir tf_objectdetect
cd tf_objectdetect/

git clone https://github.com/tensorflow/models

Download/unzip the pre-trained model in this folder

wget http://download.tensorflow.org/models/object_detection/ssdlite_mobilenet_v2_coco_2018_05_09.tar.gz
tar -xzvf ssdlite_mobilenet_v2_coco_2018_05_09.tar.gz

Copy this file models\research\object_detection\data\mscoco_label_map.pbtxt to data folder.

mkdir data
cp models/research/object_detection/data/mscoco_label_map.pbtxt data/.

compile the protos database

cd models/research/
protoc object_detection/protos/*.proto --python_out=.

cd models/research/
export PYTHONPATH=$PYTHONPATH:`pwd`:`pwd`/slim

It will be better that add this line to .bashrc

# for tensorflow object detection 
export PYTHONPATH=$PYTHONPATH:/home/pi/tf_objectdetect/models/research:/home/pi/tf_objectdetect/models/research/slim

4. Do the Object detection

Download the code, and test

cd ~/tf_objectdetect
wget https://raw.githubusercontent.com/EdjeElectronics/TensorFlow-Object-Detection-on-the-Raspberry-Pi/master/Object_detection_picamera.py

python3 Object_detection_picamera.py

You may got problems just like me, do some modify would be solved as below process.
nano Object_detection_picamera.py
Line 52

from object_detection.utils import label_map_util
from object_detection.utils import visualization_utils as vis_util

Line 66

PATH_TO_LABELS = os.path.join(CWD_PATH,'models/research/object_detection/data','mscoco_label_map.pbtxt')

That is and you will be ok for this practise

​