YOLOv5 Oriented Bounding Boxes

Overview

YOLOv5-OBB is a modified version of YOLOv5 that can predicted oriented (rotated) bounding boxes. This allows it to more closely fit oblong shapes.

Format Description

Below, learn the structure of YOLOv5 Oriented Bounding Boxes.

Each image has one txt file with a single line for each bounding box. The format of each row is

x1 y1 x2 y2 x3 y3 x4 y4 label

where fields are space delimited, and the coordinates are measured in pixels.

236 267 271 267 271 306 236 306 helmet 0
188 230 216 230 216 257 188 257 helmet 0
169 146 200 146 200 174 169 174 helmet 0
201 114 230 114 230 140 201 140 helmet 0
247 245 272 245 272 268 247 268 helmet 0
261 231 285 231 285 257 261 257 helmet 0
146 150 172 150 172 179 146 179 head 0

The `data.yaml` file contains configuration values used by the model to locate images and map class names to class_id's.

path: ../datasets/roboflow

train: train/images
val: valid/images
test: test/images

nc: 3
names: ['head', 'helmet', 'person']