Leverage GStreamer's flexibility and scalability with pre-trained deep learning models and achieve high performance AI. GstInference is an ongoing open-source project from RidgeRun that allows easy integration of state-of-the-art deep learning inference into your existing pipeline.
Latest version v0.6.0 includes:
- Supported architectures
- GoogleNet's Inception V1, V2, V3 and V4 classification architecture
- TinyYolo V2 and V3 detection architecture
- FaceNet V1 recognition architecture
- ResNet50 V1 classification architecture
- MobileNet V2 classification architecture
- Graphic debugging overlay elements
- Supported backends:
- TensorFlow (CPU)
- TensorFlow (GPU)
- NCSDK
- Supported architectures:
- x86
- Jetson Xavier, TX2, TX1, Nano
- i.MX8
Learn more in our developer's wiki:
https://developer.ridgerun.com/wiki/index.php?title=GstInference
Monitor the project progress in the project's GitHub repository: https://github.com/RidgeRun/gst-inference