Restricted Zone Access Detection with Kafka Connection

Deepstream application to detect people and cars inside pre-defined restricted zones. In case an object is detected inside these zones the bounding boxes change in color and a metadata message is send to a kafka broker with all the information of the event. For an explanation of the application and code please visit the NVIDIA Deepstream Kafka People Vehicle Detection Application medium article.

Setup Apache Kafka

The following instructions are based on the official documentation of Apache Kafka taken from here: https://kafka.apache.org/quickstart, for more details or other ways to start the service checkout the official website.

1. Download the latest kafka and extract it

tar -xzf kafka_2.13-3.4.0.tgz
cd kafka_2.13-3.4.0

2. Start the ZooKeper server in one terminal

bin/zookeeper-server-start.sh config/zookeeper.properties

3. Start the Kafka server in another terminal

bin/kafka-server-start.sh config/server.properties

4. Create a topic in another terminal

bin/kafka-topics.sh --create --topic deepstream-topic --bootstrap-server localhost:9092

5. In the same terminal run the console consumer client to read the events

bin/kafka-console-consumer.sh --topic deepstream-topic --from-beginning --bootstrap-server localhost:9092

Take into account that the the host, port and topic used to setup the server should be used at the moment of providing the connection string to the deepstream application: ;;.

How to run the application

1. Pull deepstream python bindings docker

docker pull serge3006/deepstream-python-bindings-6.1

2. Allow external applications to connect to the host's X display:

xhost +

3. Set env variables

export DISPLAY =:1

4. Run the docker container

docker run --gpus all -it --net=host --privileged -v /tmp/.X11-unix:/tmp/.X11-unix -v <app_folder>/:/app/ -w /app/ -e DISPLAY=$DISPLAY serge3006/deepstream-python-bindings-6.1

5. Install Apache Kafka requirements inside docker

The Kafka adapter uses librdkafka for the underlying protocol implementation. This library must be installed prior to use. To install librdkakfa, enter these commands:

git clone https://github.com/edenhill/librdkafka.git
cd librdkafka
git reset --hard 063a9ae7a65cebdf1cc128da9815c05f91a2a996
./configure
make
sudo make install
sudo cp /usr/local/lib/librdkafka* /opt/nvidia/deepstream/deepstream-5.0/lib

Install additional dependencies:

sudo apt-get install libglib2.0 libglib2.0-dev
sudo apt-get install libjansson4 libjansson-dev

6. Install requirements.txt

pip install -r requirements.txt

7. Run the application

python3 -m main --config_path configs/app_config.json --protolib_path /opt/nvidia/deepstream/deepstream-6.1/lib/libnvds_kafka_proto.so --connection_string localhost;9092;deepstream-topic