video-fragment-ingest 0.0.0a4

Kafka-based ingest and decoding classes for VideoIngestFragments

Video Fragment Ingest

Cloud Streaming Interfaces.

Install

pip3 install video-fragment-ingest

Api Reference Docs

https://vlt-ai.github.io/thrud/helper_classes/decoded_frame.html

Content

1. Intro
2. Usage
3. Dependencies

Intro

Consume Streaming VideoIngestFragment messages and hardware decode them into cpu np.ndarray.

message VideoIngestFragment {
  string customer_id = 1;
  string facility_id = 2;
  string camera_id = 3;
  string s3_uri = 4;
  float duration = 5;
  google.protobuf.Timestamp start_ts = 7;
  map<string, string> tags = 8;
}

Usage (Doc on private repo)

Example

Run example

Build dev container:

$./
docker build -f example/consumer/dev.Dockerfile . -t thrud.local

Enter the container:

$./
bash example/consumer/run_dev_container.sh

Once inside the container:

python3 example/consumer/sample_app.py

see python3 example/consumer/sample_app.py -h

Run a mock producer:

cd example/producer/
docker compose up -d --build

Dependencies

To run this package the environment must:

1. Have an available nvidia gpu device.
2. DEEPSTREAM_VER="7.1"
3. GSTREAMER_VER="1.24.12"