When YOLOv4 was ported to PyTorch, they decided to use the same annotation format as the Keras implementation of YOLOv3. Because the naming convention was a bit unclear, Roboflow decided that even though the underlying format definition was the same, we would keep the namespaces separate to avoid confusion about which format to use.
Note: this model has been deprecated now that YOLOv5 (PyTorch) has been released. We do not recommend using it anymore. Use YOLOv5 or YOLOv4 Darknet.
Below, learn the structure of YOLOv4 PyTorch TXT.
000049.jpg 128,168,233,275,1 158,475,290,601,1
000080.jpg 45,2,130,87,1 324,29,396,110,1 269,56,325,120,1 120,12,178,82,1 31,36,60,72,1 315,15,373,66,1 454,1,489,113,1
000038.jpg 43,46,100,108,1 244,51,304,112,1
000025.jpg 64,19,180,140,1 204,75,286,177,1 340,0,468,138,1
000087.jpg 228,41,292,108,1
000026.jpg 95,0,167,63,1 167,45,231,123,1
000068.jpg 179,4,243,88,1 40,85,76,122,1 69,64,115,108,1 2,110,29,144,1
000018.jpg 128,38,149,61,1
000037.jpg 47,22,202,199,1
000020.jpg 95,32,123,65,1 149,93,176,120,1
head
helmet
person
With Roboflow supervision, an open source Python package with utilities for completing computer vision tasks, you can merge and split detections in YOLOv4 PyTorch TXT. Read our dedicated guides to learn how to merge and split YOLOv4 PyTorch TXT detections.
Below, see model architectures that require data in the YOLOv4 PyTorch TXT format when training a new model.
On each page below, you can find links to our guides that show how to plot predictions from the model, and complete other common tasks like detecting small objects with the model.