Вернуться к статье
Распознавание опасных действий и нарушений правил использования средств индивидуальной защиты с помощью систем компьютерного зрения из видеотрансляции
Composition of the dataset: images and annotated object instances by class
as a rule an image contains objects of several classes at once, so the image counts do not sum to the size of the set
Class | Images with the class | Instances, training | Instances, validation | Share of instances, % |
Worker (worker) | 962 | 1684 | 428 | 37.0 |
Helmet (helmet) | 874 | 1198 | 306 | 26.3 |
Vest (vest) | 843 | 1114 | 283 | 24.5 |
Truck (kamaz) | 517 | 414 | 103 | 9.1 |
Climbing without a ladder (no_stairs) | 178 | 142 | 36 | 3.1 |
Total | 1000 | 4552 | 1156 | 100.0 |