yolo_single: Object Detection using YOLO

In georoen/wuepix: Using Computer Vision to Count Pedestrians

Description

detect people using YOLO+CNN (Linux C++), in a single image.

Usage

yolo_single(img, logfile = "yolo_detections.txt",
  predictions = "YOLO_Predictions/", thresh = 0.25, append = TRUE)

Arguments

img	file path to image, also known as 'now'.
logfile	file path to where to store detailed list of classification results.
predictions	dir path to where to store prediction images
thresh	numeric threshold value.
append	logical. By default (TRUE), the output is appended to the logfile. If FALSE, any existing file of the name is destroyed.

Details

depends on a working YOLO installation! See 'yolo_install()' and rerun 'yolo_update()' after updating this package (Places yolo.inst in package directory)

single processing allows storing 'predictions' (images with bounding boxes). Since these can be very insightful, you might want to 'sapply()' this function instead of 'yolo_list()'. However because then the wights have to be loaded repetitively (~10 seconds) this slows down processing.

it's recommended avoid spaces in the paths (also in working directory).

an idea for further work on this package would be to actually wrap YOLO into R (e.g. using Rccp).

Value

numeric number of detected persons.

Author(s)

Jeroen Staab

References

\insertRef

redmon2016yolo9000wuepix https://pjreddie.com/darknet/yolo/

Examples

yolo_single(img)
sapply(img.list, yolo_single)

georoen/wuepix documentation built on May 25, 2019, 5:25 p.m.