search

WongKinYiu / yolov7

Implementation of paper - YOLOv7: Trainable bag-of-freebies sets new state-of-the-art for real-time object detectors
GNU General Public License v3.0
13.38k stars 4.22k forks source link

Is MSCOCO dataset diffrent from coco dataset in this code? #151

Open 15114842111 opened 2 years ago

15114842111 commented 2 years ago

Is MSCOCO dataset diffrent from coco dataset in this code? If I just download coco dataset,can I run this code either? And how to make my custom dataset? I have one test dataset(all png), one train dataset (all png), one test_annotation.json and one train annotation.

travishsu commented 2 years ago

AFAIK, YOLO format is different from COCO-format.

Data folder in YOLO-format has following structure:

dataset
  images
    train
      1.jpg
      2.jpg
    val
      3.jpg
      4.jpg
  labels
    train
      1.txt
      2.txt
    val
      3.txt
      4.txt
  train.txt
  val.txt

Txt-file in labels may contain multiple lines, each line represent a bounding-box <class_id> <bbox_x_center> <bbox_y_center> <bbox_width> <bbox_height>

And train.txt and val.txt has recorded the paths of images