MultiDNN

Run multiple neural networks in parallel with a tiled display.

By Laurent Itti

itti@usc.edu

http://jevois.org

GPL v3

Language: C++

Supports mappings with USB output: No

Supports mappings with NO USB output: Yes

Video Mapping: NONE 0 0 0.0 YUYV 640 480 15.0 JeVois MultiDNN

Video Mapping: YUYV 640 498 15.0 YUYV 640 480 15.0 JeVois MultiDNN

Module Documentation

See DNN for more details about the JeVois DNN framework.

Edit the module's params.cfg file to select which models to run.

You can load any model at runtime by setting the pipe parameter for each pipeline.

Parameter	Type	Description	Default	Valid Values
(MultiDNN) grid	cv::Size	Grid of networks to run and display	cv::Size(2, 2)	-

params.cfg file

# Select width and height of grid:
grid=2 2

# Pipelines are named p0 (top left corner), p1, p2, ... in a horizontal raster scan on screen. For each pipeline, you
# must first set its zoo parameter to a model zoo file (default: models.yml), otherwise it will be left
# unconfigured. Then, select a pipeline from that zoo file by setting the corresponding pipe parameter:

p0:zoo=models.yml
p0:pipe=OpenCV:YuNet:YuNet-Face-512x288
p0:processing=Async

p1:zoo=models.yml
p1:pipe=NPU:Detect:yolov8n-seg-512x288
p1:processing=Async

p2:zoo=models.yml
p2:pipe=NPU:Detect:yolo11n-512x288
p2:processing=Async

p3:zoo=models.yml
p3:pipe=NPU:Pose:yolov8n-pose-512x288
p3:processing=Async

# or, if you have a Myriad-X:
#p3:pipe=VPU:Detect:pedestrian-and-vehicle-detector-adas-0001

Detailed docs:	MultiDNN
Copyright:	Copyright (C) 2018 by Laurent Itti, iLab and the University of Southern California
License:	GPL v3
Distribution:	Unrestricted
Restrictions:	None
Support URL:	http://jevois.org/doc
Other URL:	http://iLab.usc.edu
Address:	University of Southern California, HNB-07A, 3641 Watt Way, Los Angeles, CA 90089-2520, USA