GstInference/Overlay Elements

GstInference
Introduction
Getting started
Getting the code Building the plugin Quick Starting Guide
Supported architectures
InceptionV1 InceptionV2 InceptionV3 InceptionV4 TinyYoloV2 TinyYoloV3 YoloV2 ResNetV1 MobileNetV2 MobileNetV2 + SSD AlexNet
Supported backends
TensorFlow Tensorflow-Lite Coral from Google Caffe TensorRT ONNXRT ONNXRT ACL ONNXRT OpenVINO
Metadata and Signals
GstClassificationMeta GstInferenceMeta Signals
Overlay Elements
Inference Overlay
Utils Elements
InferenceFilter InferenceCrop InferenceDebug InferenceBin
Legacy pipelines
PC NANO TX2 Xavier IMX8
Example pipelines
Example applications
Classification Detection Hand Pose Recognition
Benchmarks
Model Zoo
Project Status
Releases Roadmap
Contact Us

Alongside GstInference's inference plugins, we provide a series of elements that help visualize inference results on GStreamer. This section provides documentation on these elements.

Element	Description	Status
classificationoverlay	Creates a text overlay with the most probable class on an incoming buffer with classification metadata	Complete
detectionoverlay	Draws a square with its respective label for every bounding box contained in the buffer's detection metadata	Complete
embeddingoverlay	Draws a green border and writes "Pass" on the frame if its embedding is close to a known value. Draws a red border and writes "Fail" otherwise.	Complete
inferenceoverlay	Draws a square with its respective label for every bounding box contained in the buffer's inference metadata	Complete
segmentationoverlay	-	Coming soon

Installation

To build the overlay elements you need OpenCV >= 3.3.1. Enable or disable the construction of the overlay elements using the option --<enable/disable>-opencv during autogen or configure:

./autogen.sh --enable-opencv #Build the elements (default)
./autogen.sh --disable-opencv #Don't build the elements