Explore Savant Demos

Browse our selection of samples and demos highlighting the capabilities of our computer vision framework. They provide a practical starting point for an understanding of the Savant technology.
Get started at your own pace: samples are organized from introductory to advanced.
Person Counting
Demonstration of how to count objects located in the defined polygonal areas with the PeopleNet detection model and Savant.
RT-DETR Object Detection
Demonstration of how to use the innovative RT-DETR model with Savant
Person Detection & Anonymization
The demo shows the use of PeopleNet and OpenCV CUDA to detect people and their faces and modify the video stream to anonymize persons.
Car Classification
The sample shows how to use the models from Nvidia Zoo to detect and classify cars.
Traffic Metering
The demo is based on YOLOV8 or PeopleNet, Graphite, Grafana
Advanced Road Traffic Metering
Road traffic metering by directions with YOLOV8
Car License Plate Recognition
NGC-based car license plate recognition
KeyPoint Detection
YOLOV8-KeyPoint Sample
Instance Segmentation
YOLOV8-Seg Sample
Age/Gender Prediction
YOLOV5-Face, Custom Age/Gender Model
Facial Re-ID
YOLOV5-Face, AdaFace and HNSWLIB
AnimeGANv2
The demo shows the use of generative neural networks with Savant
Video Upscaling / Super-Resolution
The demo shows how to enhance video (x3 upscale) in real-time with a super resolution model
PyTorch Serving
The sample demonstrates how to serve a PyTorch inference and postprocessing code with Savant
Simultaneous Multi-resolution Video Transcoding And RTSP Broadcasting
Demo, showing how to use Savant to simultaneously transcode multiple incoming streams to outgoing streams in several resolutions like FHD, HD, SD, etc.
No content