Description

• simple object detection in specific area (polygon) for models from TF API Model ZOO

Installation

python3 -m venv venv
source venv/bin/activate

pip install -U wheel pip setuptools
pip install -r requirements.txt

# TF installation
pip install tensorflow==1.15  # for CPU
# or
pip install tensorflow-gpu==1.15  # for CPU | GPU

# pull models
export GOOGLE_APPLICATION_CREDENTIALS=$PWD/credentials/gs_viewer.json
dvc pull

Usage

export TF_CPP_MIN_LOG_LEVEL=5  # to avoid TF logs


# just object detection
python detect.py -s data/images/normal.jpg -c configs/tf_object_api_cfg.yml -p "[0,0], [0,1], [1,1], [1,0]" -a 0.0001

# with masks
python detect.py -s data/images/normal.jpg -c configs/tf_object_api_cfg_masks.yml -p "[0,0], [0,1], [1,1], [1,0]" -a 0.0001

• Note: model's config explained

Demo

./demo.sh

# check ./demo folder