DNN Vision Library (jetson-inference)

C++ runtime library for supporting deep vision neural network models and inferencing with TensorRT. Available network primitives currently include:
More...

Modules
	tensorNet
	DNN abstract base class that provides TensorRT functionality underneath. These functions aren't typically accessed by end users unless they are implementing their own DNN class like imageNet or detectNet.
	imageNet
	Image recognition DNN (GoogleNet, AlexNet, ResNet)
	detectNet
	Object detection DNN (SSD, DetectNet)
	segNet
	Semantic segmentation DNN (FCN or Fully-Convolutional Networks)
	poseNet
	Pose estimation DNN.
	actionNet
	Action/activity recognition DNN.
	depthNet
	Mono depth estimation from monocular images.
	backgroundNet
	Foreground/background segmentation and removal DNN.
	Object Tracking
	Object tracking used by detectNet.
	Model Downloader
	Utilities for automatically downloading pre-trained models.

Detailed Description

C++ runtime library for supporting deep vision neural network models and inferencing with TensorRT. Available network primitives currently include:

• imageNet for image recognition
• detectNet for object detection + localization
• segNet for semantic segmentation
• poseNet for pose estimation
• depthNet for mono depth

The different primitives each inherit from the shared tensorNet object.

See also

Utilities Library (jetson-utils) for various utilities, including CUDA functions and Video Streaming interfaces.