tflite_obj_detection

TF lite model to count number of humans in a space.

usage

• Get the model and its labels. Takes argument of the path, if blank saves in ./models directory

bash ./download.sh ./models

• Run the script

python3 model.py \
        --model ./models/detect.tflite \
        --labels ./models/coco_labels.txt \
        --image ./bk.png

Setup on raspberry PI

• If you're running Debian Linux or a derivative of Debian (including Raspberry Pi OS), you should install from our Debian package repo. This requires that you add a new repo list and key to your system and then install as follows:

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 -
sudo apt-get update
sudo apt-get install python3-tflite-runtime

• With pip Recommended

pip3 install --index-url https://google-coral.github.io/py-repo/ tflite_runtime

• There are sometimes issues reported on Raspberry Pi setups when installing numpy using pip3 install (or pip install). Run this:

sudo apt-get install libatlas-base-dev

• Alternatively use the NumPy provided by Raspbian. In which case run:

pip3 uninstall numpy  # remove previously installed version
apt install python3-numpy