Cattle detection and counting in UAV images
based on convolutional neural networks

Wen Shao1, Rei Kawakami1, Ryota Yoshihashi1, Shaodi You2, Hidemichi Kawase3, Takeshi Naemura1
1The University of Tokyo, 2Data61-CSIRO, 3Kamiens Technology Inc.
an image in the dataset

Publication

Cattle detection and counting in UAV images based on convolutional neural networks

Wen Shao, Rei Kawakami, Ryota Yoshihashi, Shaodi You, Hidemichi Kawase, Takeshi Naemura
International Journal of Remote Sensing, 41:1, 31-52, DOI: 10.1080/01431161.2019.1624858 (2020)

Download

Usage

The annotation txt file have the following format for each image:
image num x1 y1 w1 h1 quality id id_conf x2 y2 w2 h2 quality id id_conf ...
In this file, items are separated by tabs.
"image" means the name of image file.
"num" means the number of annotations in this image.
"x1,y1,w1,h1" means a bounding box whose left-top point is (x1, y1), width is w1 and height is h1.
The quality label -- "0": Normal, "1": Truncated, "2": Blurred, "3": Occluded
The cattle id -- In each flight, the same target is labelled with the same id.
The "id_conf" -- "0": lack of confidence when labelling, "1": high confidence when labelling