์ฝ˜ํ…์ธ ๋กœ ๊ฑด๋„ˆ๋›ฐ๊ธฐ

๊ฐ์ฒด ๊ฐ์ง€ ๋ฐ์ดํ„ฐ ์„ธํŠธ ๊ฐœ์š”

๊ฐ•๋ ฅํ•˜๊ณ  ์ •ํ™•ํ•œ ๊ฐ์ฒด ๊ฐ์ง€ ๋ชจ๋ธ์„ ํ›ˆ๋ จํ•˜๋ ค๋ฉด ํฌ๊ด„์ ์ธ ๋ฐ์ดํ„ฐ ์„ธํŠธ๊ฐ€ ํ•„์š”ํ•ฉ๋‹ˆ๋‹ค. ์ด ๊ฐ€์ด๋“œ์—์„œ๋Š” Ultralytics YOLO ๋ชจ๋ธ๊ณผ ํ˜ธํ™˜๋˜๋Š” ๋‹ค์–‘ํ•œ ํ˜•์‹์˜ ๋ฐ์ดํ„ฐ์…‹์„ ์†Œ๊ฐœํ•˜๊ณ  ๊ทธ ๊ตฌ์กฐ, ์‚ฌ์šฉ๋ฒ•, ์„œ๋กœ ๋‹ค๋ฅธ ํ˜•์‹ ๊ฐ„ ๋ณ€ํ™˜ ๋ฐฉ๋ฒ•์— ๋Œ€ํ•œ ์ธ์‚ฌ์ดํŠธ๋ฅผ ์ œ๊ณตํ•ฉ๋‹ˆ๋‹ค.

์ง€์›๋˜๋Š” ๋ฐ์ดํ„ฐ ์„ธํŠธ ํ˜•์‹

Ultralytics YOLO ํ˜•์‹

Ultralytics YOLO ํ˜•์‹์€ ๋ฐ์ดํ„ฐ ์„ธํŠธ ๊ตฌ์„ฑ ํ˜•์‹์œผ๋กœ, ๋ฐ์ดํ„ฐ ์„ธํŠธ ๋ฃจํŠธ ๋””๋ ‰ํ„ฐ๋ฆฌ, ํŠธ๋ ˆ์ด๋‹/๊ฒ€์ฆ/ํ…Œ์ŠคํŠธ ์ด๋ฏธ์ง€ ๋””๋ ‰ํ„ฐ๋ฆฌ์˜ ์ƒ๋Œ€ ๊ฒฝ๋กœ ๋˜๋Š” *.txt ํŒŒ์ผ๊ณผ ์ด๋ฏธ์ง€ ๊ฒฝ๋กœ๊ฐ€ ํฌํ•จ๋œ ํด๋ž˜์Šค ์ด๋ฆ„ ์‚ฌ์ „์ด ์žˆ์Šต๋‹ˆ๋‹ค. ๋‹ค์Œ์€ ์˜ˆ์‹œ์ž…๋‹ˆ๋‹ค:

# Train/val/test sets as 1) dir: path/to/imgs, 2) file: path/to/imgs.txt, or 3) list: [path/to/imgs1, path/to/imgs2, ..]
path: ../datasets/coco8 # dataset root dir
train: images/train # train images (relative to 'path') 4 images
val: images/val # val images (relative to 'path') 4 images
test: # test images (optional)

# Classes (80 COCO classes)
names:
    0: person
    1: bicycle
    2: car
    # ...
    77: teddy bear
    78: hair drier
    79: toothbrush

์ด ํ˜•์‹์˜ ๋ ˆ์ด๋ธ”์€ YOLO ํ˜•์‹์œผ๋กœ ๋‚ด๋ณด๋‚ด์•ผ ํ•ฉ๋‹ˆ๋‹ค. *.txt ํŒŒ์ผ์„ ์ถ”๊ฐ€ํ•ฉ๋‹ˆ๋‹ค. ์ด๋ฏธ์ง€์— ๊ฐ์ฒด๊ฐ€ ์—†๋Š” ๊ฒฝ์šฐ์—๋Š” *.txt ํŒŒ์ผ์ด ํ•„์š”ํ•ฉ๋‹ˆ๋‹ค. ํŒŒ์ผ์€ *.txt ํŒŒ์ผ์€ ๊ฐ์ฒด๋‹น ํ•œ ํ–‰์œผ๋กœ ํฌ๋งทํ•ด์•ผ ํ•ฉ๋‹ˆ๋‹ค. class x_center y_center width height ํ˜•์‹์ด์–ด์•ผ ํ•ฉ๋‹ˆ๋‹ค. ์ƒ์ž ์ขŒํ‘œ๋Š” ์ •๊ทœํ™”๋œ xywh ํ˜•์‹(0์—์„œ 1๊นŒ์ง€)์„ ์‚ฌ์šฉํ•ด์•ผ ํ•ฉ๋‹ˆ๋‹ค. ์ƒ์ž๊ฐ€ ํ”ฝ์…€ ๋‹จ์œ„์ธ ๊ฒฝ์šฐ ๋‹ค์Œ์„ ๋‚˜๋ˆ„์–ด์•ผ ํ•ฉ๋‹ˆ๋‹ค. x_center ๊ทธ๋ฆฌ๊ณ  width ๋ฅผ ์ด๋ฏธ์ง€ ๋„ˆ๋น„๋กœ, ๊ทธ๋ฆฌ๊ณ  y_center ๊ทธ๋ฆฌ๊ณ  height ๋ฅผ ์ด๋ฏธ์ง€ ๋†’์ด๋ณ„๋กœ ์ •๋ ฌํ•ฉ๋‹ˆ๋‹ค. ํด๋ž˜์Šค ๋ฒˆํ˜ธ๋Š” 0์œผ๋กœ ์‹œ์ž‘ํ•˜๋Š” ์˜ ์ธ๋ฑ์Šค์—ฌ์•ผ ํ•ฉ๋‹ˆ๋‹ค.

๋ ˆ์ด๋ธ”์ด ์ง€์ •๋œ ์ด๋ฏธ์ง€ ์˜ˆ์‹œ

์œ„ ์ด๋ฏธ์ง€์— ํ•ด๋‹นํ•˜๋Š” ๋ผ๋ฒจ ํŒŒ์ผ์—๋Š” 2์ธ(ํด๋ž˜์Šค 0) ๋ฐ ๋™์ (ํด๋ž˜์Šค 27):

๋ ˆ์ด๋ธ” ํŒŒ์ผ ์˜ˆ์‹œ

Ultralytics YOLO ํ˜•์‹์„ ์‚ฌ์šฉํ•˜๋Š” ๊ฒฝ์šฐ ์•„๋ž˜ COCO8 ๋ฐ์ดํ„ฐ ์„ธํŠธ ์˜ˆ์‹œ์™€ ๊ฐ™์ด ํŠธ๋ ˆ์ด๋‹ ๋ฐ ์œ ํšจ์„ฑ ๊ฒ€์‚ฌ ์ด๋ฏธ์ง€์™€ ๋ ˆ์ด๋ธ”์„ ๊ตฌ์„ฑํ•ฉ๋‹ˆ๋‹ค.

๋ฐ์ดํ„ฐ ์„ธํŠธ ๋””๋ ‰ํ„ฐ๋ฆฌ ๊ตฌ์กฐ ์˜ˆ์‹œ

์‚ฌ์šฉ๋ฒ•

์ด๋Ÿฌํ•œ ํ˜•์‹์„ ์‚ฌ์šฉํ•˜์—ฌ ๋ชจ๋ธ์„ ํ›ˆ๋ จํ•˜๋Š” ๋ฐฉ๋ฒ•์€ ๋‹ค์Œ๊ณผ ๊ฐ™์Šต๋‹ˆ๋‹ค:

์˜ˆ

from ultralytics import YOLO

# Load a model
model = YOLO("yolo11n.pt")  # load a pretrained model (recommended for training)

# Train the model
results = model.train(data="coco8.yaml", epochs=100, imgsz=640)
# Start training from a pretrained *.pt model
yolo detect train data=coco8.yaml model=yolo11n.pt epochs=100 imgsz=640

์ง€์›๋˜๋Š” ๋ฐ์ดํ„ฐ ์„ธํŠธ

๋‹ค์Œ์€ ์ง€์›๋˜๋Š” ๋ฐ์ดํ„ฐ ์„ธํŠธ ๋ชฉ๋ก๊ณผ ๊ฐ ๋ฐ์ดํ„ฐ ์„ธํŠธ์— ๋Œ€ํ•œ ๊ฐ„๋žตํ•œ ์„ค๋ช…์ž…๋‹ˆ๋‹ค:

  • Argoverse: ํ’๋ถ€ํ•œ ์ฃผ์„์ด ํฌํ•จ๋œ ๋„์‹œ ํ™˜๊ฒฝ์˜ 3D ์ถ”์  ๋ฐ ๋ชจ์…˜ ์˜ˆ์ธก ๋ฐ์ดํ„ฐ๊ฐ€ ํฌํ•จ๋œ ๋ฐ์ดํ„ฐ ์„ธํŠธ์ž…๋‹ˆ๋‹ค.
  • COCO: ์ปจํ…์ŠคํŠธ ๋‚ด ๊ณตํ†ต ๊ฐœ์ฒด(COCO)๋Š” 80๊ฐœ์˜ ๊ฐœ์ฒด ๋ฒ”์ฃผ๋กœ ๊ตฌ์„ฑ๋œ ๋Œ€๊ทœ๋ชจ ๊ฐœ์ฒด ๊ฐ์ง€, ์„ธ๋ถ„ํ™” ๋ฐ ์บก์…˜ ๋ฐ์ดํ„ฐ ์„ธํŠธ์ž…๋‹ˆ๋‹ค.
  • LVIS: 1203๊ฐœ์˜ ๊ฐ์ฒด ์นดํ…Œ๊ณ ๋ฆฌ๊ฐ€ ํฌํ•จ๋œ ๋Œ€๊ทœ๋ชจ ๊ฐ์ฒด ๊ฐ์ง€, ์„ธ๋ถ„ํ™” ๋ฐ ์บก์…˜ ๋ฐ์ดํ„ฐ ์„ธํŠธ์ž…๋‹ˆ๋‹ค.
  • COCO8: ๋น ๋ฅธ ํ…Œ์ŠคํŠธ์— ์ ํ•ฉํ•œ COCO train ๋ฐ COCO val์˜ ์ฒ˜์Œ 4๊ฐœ ์ด๋ฏธ์ง€์˜ ์ž‘์€ ํ•˜์œ„ ์ง‘ํ•ฉ์ž…๋‹ˆ๋‹ค.
  • COCO128: ํ…Œ์ŠคํŠธ์— ์ ํ•ฉํ•œ COCO train ๋ฐ COCO val์˜ ์ฒซ 128๊ฐœ ์ด๋ฏธ์ง€์˜ ์ž‘์€ ํ•˜์œ„ ์ง‘ํ•ฉ์ž…๋‹ˆ๋‹ค.
  • ๊ธ€๋กœ๋ฒŒ ๋ฐ€ 2020: ๊ธ€๋กœ๋ฒŒ ๋ฐ€ ์ฑŒ๋ฆฐ์ง€ 2020์˜ ๋ฐ€ ๋จธ๋ฆฌ ์ด๋ฏธ์ง€๊ฐ€ ํฌํ•จ๋œ ๋ฐ์ดํ„ฐ ์„ธํŠธ์ž…๋‹ˆ๋‹ค.
  • Objects365: 365๊ฐœ์˜ ๋ฌผ์ฒด ์นดํ…Œ๊ณ ๋ฆฌ์™€ 60๋งŒ ๊ฐœ ์ด์ƒ์˜ ์ฃผ์„์ด ๋‹ฌ๋ฆฐ ์ด๋ฏธ์ง€๊ฐ€ ํฌํ•จ๋œ ๊ณ ํ’ˆ์งˆ์˜ ๋Œ€๊ทœ๋ชจ ๋ฌผ์ฒด ๊ฐ์ง€ ๋ฐ์ดํ„ฐ ์„ธํŠธ์ž…๋‹ˆ๋‹ค.
  • OpenImagesV7: 170๋งŒ ๊ฐœ์˜ ์—ด์ฐจ ์ด๋ฏธ์ง€์™€ 4๋งŒ 2์ฒœ ๊ฐœ์˜ ์œ ํšจ์„ฑ ๊ฒ€์‚ฌ ์ด๋ฏธ์ง€๊ฐ€ ํฌํ•จ๋œ Google ์˜ ํฌ๊ด„์ ์ธ ๋ฐ์ดํ„ฐ ์„ธํŠธ์ž…๋‹ˆ๋‹ค.
  • SKU-110K: 11K ๊ฐœ ์ด์ƒ์˜ ์ด๋ฏธ์ง€์™€ 170๋งŒ ๊ฐœ์˜ ๊ฒฝ๊ณ„ ์ƒ์ž๋กœ ๊ตฌ์„ฑ๋œ ์†Œ๋งค ํ™˜๊ฒฝ์—์„œ์˜ ๊ณ ๋ฐ€๋„ ๋ฌผ์ฒด ๊ฐ์ง€ ๊ธฐ๋Šฅ์„ ๊ฐ–์ถ˜ ๋ฐ์ดํ„ฐ ์„ธํŠธ์ž…๋‹ˆ๋‹ค.
  • VisDrone: 10K ์ด์ƒ์˜ ์ด๋ฏธ์ง€์™€ ๋น„๋””์˜ค ์‹œํ€€์Šค๊ฐ€ ํฌํ•จ๋œ ๋“œ๋ก ์œผ๋กœ ์บก์ฒ˜ํ•œ ์ด๋ฏธ์ง€์˜ ๊ฐ์ฒด ๊ฐ์ง€ ๋ฐ ๋‹ค์ค‘ ๊ฐ์ฒด ์ถ”์  ๋ฐ์ดํ„ฐ๊ฐ€ ํฌํ•จ๋œ ๋ฐ์ดํ„ฐ ์„ธํŠธ์ž…๋‹ˆ๋‹ค.
  • VOC: 20๊ฐœ์˜ ๊ฐ์ฒด ํด๋ž˜์Šค์™€ 11,000๊ฐœ ์ด์ƒ์˜ ์ด๋ฏธ์ง€๊ฐ€ ํฌํ•จ๋œ ๊ฐ์ฒด ๊ฐ์ง€ ๋ฐ ๋ถ„ํ• ์„ ์œ„ํ•œ ํŒŒ์Šค์นผ ์‹œ๊ฐ ๊ฐ์ฒด ํด๋ž˜์Šค(VOC) ๋ฐ์ดํ„ฐ ์„ธํŠธ์ž…๋‹ˆ๋‹ค.
  • xView: 60๊ฐœ์˜ ๊ฐ์ฒด ์นดํ…Œ๊ณ ๋ฆฌ์™€ 100๋งŒ ๊ฐœ ์ด์ƒ์˜ ์ฃผ์„์ด ๋‹ฌ๋ฆฐ ์˜ค๋ฒ„ํ—ค๋“œ ์ด๋ฏธ์ง€์—์„œ ๊ฐ์ฒด๋ฅผ ๊ฐ์ง€ํ•˜๊ธฐ ์œ„ํ•œ ๋ฐ์ดํ„ฐ ์„ธํŠธ์ž…๋‹ˆ๋‹ค.
  • Roboflow 100: ํฌ๊ด„์ ์ธ ๋ชจ๋ธ ํ‰๊ฐ€๋ฅผ ์œ„ํ•ด 7๊ฐœ ์ด๋ฏธ์ง€ ๋„๋ฉ”์ธ์— ๊ฑธ์ณ 100๊ฐœ์˜ ๋ฐ์ดํ„ฐ ์„ธํŠธ๋ฅผ ํฌํ•จํ•˜๋Š” ๋‹ค์–‘ํ•œ ๋ฌผ์ฒด ๊ฐ์ง€ ๋ฒค์น˜๋งˆํฌ์ž…๋‹ˆ๋‹ค.
  • ๋‡Œ์ข…์–‘: ๋‡Œ์ข…์–‘์„ ๊ฐ์ง€ํ•˜๊ธฐ ์œ„ํ•œ ๋ฐ์ดํ„ฐ ์„ธํŠธ์—๋Š” ์ข…์–‘์˜ ์กด์žฌ ์—ฌ๋ถ€, ์œ„์น˜ ๋ฐ ํŠน์„ฑ์— ๋Œ€ํ•œ ์„ธ๋ถ€ ์ •๋ณด๊ฐ€ ํฌํ•จ๋œ MRI ๋˜๋Š” CT ์Šค์บ” ์ด๋ฏธ์ง€๊ฐ€ ํฌํ•จ๋ฉ๋‹ˆ๋‹ค.
  • ์•„ํ”„๋ฆฌ์นด ์•ผ์ƒ๋™๋ฌผ: ๋ฒ„ํŒ”๋กœ, ์ฝ”๋ผ๋ฆฌ, ์ฝ”๋ฟ”์†Œ, ์–ผ๋ฃฉ๋ง ๋“ฑ ์•„ํ”„๋ฆฌ์นด ์•ผ์ƒ๋™๋ฌผ์˜ ์ด๋ฏธ์ง€๊ฐ€ ํฌํ•จ๋œ ๋ฐ์ดํ„ฐ ์„ธํŠธ์ž…๋‹ˆ๋‹ค.
  • ์„œ๋ช…: ์„œ๋ช…: ์ฃผ์„์ด ๋‹ฌ๋ฆฐ ์„œ๋ช…์ด ์žˆ๋Š” ๋‹ค์–‘ํ•œ ๋ฌธ์„œ ์ด๋ฏธ์ง€๋กœ ๊ตฌ์„ฑ๋œ ๋ฐ์ดํ„ฐ ์„ธํŠธ๋กœ, ๋ฌธ์„œ ๊ฒ€์ฆ ๋ฐ ์‚ฌ๊ธฐ ํƒ์ง€ ์—ฐ๊ตฌ๋ฅผ ์ง€์›ํ•ฉ๋‹ˆ๋‹ค.

๋‚˜๋งŒ์˜ ๋ฐ์ดํ„ฐ ์ง‘ํ•ฉ ์ถ”๊ฐ€ํ•˜๊ธฐ

์ž์ฒด ๋ฐ์ดํ„ฐ ์„ธํŠธ๊ฐ€ ์žˆ๊ณ  ์ด๋ฅผ Ultralytics YOLO ํ˜•์‹์œผ๋กœ ํƒ์ง€ ๋ชจ๋ธ ํ•™์Šต์— ์‚ฌ์šฉํ•˜๋ ค๋Š” ๊ฒฝ์šฐ, ์œ„์˜ "Ultralytics YOLO ํ˜•์‹"์— ์ง€์ •๋œ ํ˜•์‹์„ ๋”ฐ๋ฅด๋Š”์ง€ ํ™•์ธํ•˜์„ธ์š”. ์–ด๋…ธํ…Œ์ด์…˜์„ ํ•„์š”ํ•œ ํ˜•์‹์œผ๋กœ ๋ณ€ํ™˜ํ•˜๊ณ  YAML ๊ตฌ์„ฑ ํŒŒ์ผ์— ๊ฒฝ๋กœ, ํด๋ž˜์Šค ์ˆ˜, ํด๋ž˜์Šค ์ด๋ฆ„์„ ์ง€์ •ํ•˜์„ธ์š”.

๋ผ๋ฒจ ํ˜•์‹ ํฌํŠธ ๋˜๋Š” ๋ณ€ํ™˜

COCO ๋ฐ์ดํ„ฐ ์„ธํŠธ ํ˜•์‹์„ YOLO ํ˜•์‹์œผ๋กœ ๋ณ€ํ™˜

๋‹ค์Œ ์ฝ”๋“œ ์Šค๋‹ˆํŽซ์„ ์‚ฌ์šฉํ•˜์—ฌ ๋„๋ฆฌ ์‚ฌ์šฉ๋˜๋Š” COCO ๋ฐ์ดํ„ฐ ์„ธํŠธ ํ˜•์‹์—์„œ YOLO ํ˜•์‹์œผ๋กœ ๋ผ๋ฒจ์„ ์‰ฝ๊ฒŒ ๋ณ€ํ™˜ํ•  ์ˆ˜ ์žˆ์Šต๋‹ˆ๋‹ค:

์˜ˆ

from ultralytics.data.converter import convert_coco

convert_coco(labels_dir="path/to/coco/annotations/")

์ด ๋ณ€ํ™˜ ๋„๊ตฌ๋Š” COCO ๋ฐ์ดํ„ฐ ์„ธํŠธ ๋˜๋Š” COCO ํ˜•์‹์˜ ๋ฐ์ดํ„ฐ ์„ธํŠธ๋ฅผ Ultralytics YOLO ํ˜•์‹์œผ๋กœ ๋ณ€ํ™˜ํ•˜๋Š” ๋ฐ ์‚ฌ์šฉํ•  ์ˆ˜ ์žˆ์Šต๋‹ˆ๋‹ค.

์‚ฌ์šฉํ•˜๋ ค๋Š” ๋ฐ์ดํ„ฐ ์„ธํŠธ๊ฐ€ ๋ชจ๋ธ๊ณผ ํ˜ธํ™˜๋˜๋Š”์ง€, ํ•„์š”ํ•œ ํ˜•์‹ ๊ทœ์น™์„ ๋”ฐ๋ฅด๊ณ  ์žˆ๋Š”์ง€ ๋‹ค์‹œ ํ•œ ๋ฒˆ ํ™•์ธํ•˜๋Š” ๊ฒƒ์„ ์žŠ์ง€ ๋งˆ์„ธ์š”. ์˜ฌ๋ฐ”๋ฅธ ํ˜•์‹์˜ ๋ฐ์ดํ„ฐ ์„ธํŠธ๋Š” ์„ฑ๊ณต์ ์ธ ๊ฐ์ฒด ๊ฐ์ง€ ๋ชจ๋ธ์„ ํ›ˆ๋ จํ•˜๋Š” ๋ฐ ๋งค์šฐ ์ค‘์š”ํ•ฉ๋‹ˆ๋‹ค.

์ž์ฃผ ๋ฌป๋Š” ์งˆ๋ฌธ

Ultralytics YOLO ๋ฐ์ดํ„ฐ ์„ธํŠธ ํ˜•์‹์ด๋ž€ ๋ฌด์—‡์ด๋ฉฐ ์–ด๋–ป๊ฒŒ ๊ตฌ์„ฑํ•˜๋‚˜์š”?

Ultralytics YOLO ํ˜•์‹์€ ํŠธ๋ ˆ์ด๋‹ ํ”„๋กœ์ ํŠธ์—์„œ ๋ฐ์ดํ„ฐ ์„ธํŠธ๋ฅผ ์ •์˜ํ•˜๊ธฐ ์œ„ํ•œ ๊ตฌ์กฐํ™”๋œ ๊ตฌ์„ฑ์ž…๋‹ˆ๋‹ค. ์—ฌ๊ธฐ์—๋Š” ํŠธ๋ ˆ์ด๋‹, ์œ ํšจ์„ฑ ๊ฒ€์‚ฌ ๋ฐ ํ…Œ์ŠคํŠธ ์ด๋ฏธ์ง€์™€ ํ•ด๋‹น ๋ ˆ์ด๋ธ”์— ๋Œ€ํ•œ ๊ฒฝ๋กœ ์„ค์ •์ด ํฌํ•จ๋ฉ๋‹ˆ๋‹ค. ์˜ˆ๋ฅผ ๋“ค์–ด

path: ../datasets/coco8 # dataset root directory
train: images/train # training images (relative to 'path')
val: images/val # validation images (relative to 'path')
test: # optional test images
names:
    0: person
    1: bicycle
    2: car
    # ...

๋ ˆ์ด๋ธ”์€ ๋‹ค์Œ ์œ„์น˜์— ์ €์žฅ๋ฉ๋‹ˆ๋‹ค. *.txt ํŒŒ์ผ์€ ์ด๋ฏธ์ง€๋‹น ํ•˜๋‚˜์˜ ํŒŒ์ผ๋กœ, ํ˜•์‹์€ class x_center y_center width height ๋ฅผ ์ •๊ทœํ™”๋œ ์ขŒํ‘œ๋กœ ๋ณ€ํ™˜ํ•ฉ๋‹ˆ๋‹ค. ์ž์„ธํ•œ ๊ฐ€์ด๋“œ๋Š” COCO8 ๋ฐ์ดํ„ฐ ์„ธํŠธ ์˜ˆ์‹œ.

COCO ๋ฐ์ดํ„ฐ์…‹์„ YOLO ํ˜•์‹์œผ๋กœ ๋ณ€ํ™˜ํ•˜๋ ค๋ฉด ์–ด๋–ป๊ฒŒ ํ•˜๋‚˜์š”?

Ultralytics ๋ณ€ํ™˜ ๋„๊ตฌ๋ฅผ ์‚ฌ์šฉํ•˜์—ฌ COCO ๋ฐ์ดํ„ฐ ์ง‘ํ•ฉ์„ YOLO ํ˜•์‹์œผ๋กœ ๋ณ€ํ™˜ํ•  ์ˆ˜ ์žˆ์Šต๋‹ˆ๋‹ค. ๋‹ค์Œ์€ ๋น ๋ฅธ ๋ฐฉ๋ฒ•์ž…๋‹ˆ๋‹ค:

from ultralytics.data.converter import convert_coco

convert_coco(labels_dir="path/to/coco/annotations/")

์ด ์ฝ”๋“œ๋Š” COCO ์ฃผ์„์„ YOLO ํ˜•์‹์œผ๋กœ ๋ณ€ํ™˜ํ•˜์—ฌ Ultralytics YOLO ๋ชจ๋ธ๊ณผ ์›ํ™œํ•˜๊ฒŒ ํ†ตํ•ฉํ•  ์ˆ˜ ์žˆ๋„๋ก ํ•ฉ๋‹ˆ๋‹ค. ์ž์„ธํ•œ ๋‚ด์šฉ์€ ํฌํŠธ ๋˜๋Š” ๋ผ๋ฒจ ํ˜•์‹ ๋ณ€ํ™˜ ์„น์…˜์„ ์ฐธ์กฐํ•˜์„ธ์š”.

๊ฐ์ฒด ๊ฐ์ง€๋ฅผ ์œ„ํ•ด Ultralytics YOLO ์—์„œ ์ง€์›ํ•˜๋Š” ๋ฐ์ดํ„ฐ ์„ธํŠธ๋Š” ๋ฌด์—‡์ธ๊ฐ€์š”?

Ultralytics YOLO ๋ฅผ ํฌํ•จํ•œ ๋‹ค์–‘ํ•œ ๋ฐ์ดํ„ฐ ์„ธํŠธ๋ฅผ ์ง€์›ํ•ฉ๋‹ˆ๋‹ค:

๊ฐ ๋ฐ์ดํ„ฐ ์„ธํŠธ ํŽ˜์ด์ง€์—์„œ๋Š” ํšจ์œจ์ ์ธ YOLO11 ๊ต์œก์„ ์œ„ํ•œ ๊ตฌ์กฐ์™€ ์‚ฌ์šฉ๋ฒ•์— ๋Œ€ํ•œ ์ž์„ธํ•œ ์ •๋ณด๋ฅผ ์ œ๊ณตํ•ฉ๋‹ˆ๋‹ค. ์ง€์›๋˜๋Š” ๋ฐ์ดํ„ฐ ์„ธํŠธ ์„น์…˜์—์„œ ์ „์ฒด ๋ชฉ๋ก์„ ์‚ดํŽด๋ณด์„ธ์š”.

๋‚ด ๋ฐ์ดํ„ฐ ์ง‘ํ•ฉ์„ ์‚ฌ์šฉํ•˜์—ฌ YOLO11 ๋ชจ๋ธ ํ•™์Šต์„ ์‹œ์ž‘ํ•˜๋ ค๋ฉด ์–ด๋–ป๊ฒŒ ํ•ด์•ผ ํ•˜๋‚˜์š”?

YOLO11 ๋ชจ๋ธ ํ•™์Šต์„ ์‹œ์ž‘ํ•˜๋ ค๋ฉด ๋ฐ์ดํ„ฐ ์„ธํŠธ์˜ ํ˜•์‹์ด ์˜ฌ๋ฐ”๋ฅด๊ฒŒ ์ง€์ •๋˜์–ด ์žˆ๊ณ  ๊ฒฝ๋กœ๊ฐ€ YAML ํŒŒ์ผ์— ์ •์˜๋˜์–ด ์žˆ๋Š”์ง€ ํ™•์ธํ•ฉ๋‹ˆ๋‹ค. ๋‹ค์Œ ์Šคํฌ๋ฆฝํŠธ๋ฅผ ์‚ฌ์šฉํ•˜์—ฌ ํ›ˆ๋ จ์„ ์‹œ์ž‘ํ•ฉ๋‹ˆ๋‹ค:

์˜ˆ

from ultralytics import YOLO

model = YOLO("yolo11n.pt")  # Load a pretrained model
results = model.train(data="path/to/your_dataset.yaml", epochs=100, imgsz=640)
yolo detect train data=path/to/your_dataset.yaml model=yolo11n.pt epochs=100 imgsz=640

CLI ๋ช…๋ น์–ด๋ฅผ ๋น„๋กฏํ•œ ๋‹ค์–‘ํ•œ ๋ชจ๋“œ ํ™œ์šฉ์— ๋Œ€ํ•œ ์ž์„ธํ•œ ๋‚ด์šฉ์€ ์‚ฌ์šฉ ์„น์…˜์„ ์ฐธ์กฐํ•˜์„ธ์š”.

๋ฌผ์ฒด ๊ฐ์ง€๋ฅผ ์œ„ํ•ด Ultralytics YOLO ์„ ์‚ฌ์šฉํ•˜๋Š” ์‹ค์ œ ์‚ฌ๋ก€๋Š” ์–ด๋””์—์„œ ์ฐพ์„ ์ˆ˜ ์žˆ๋‚˜์š”?

Ultralytics ๋Š” ๋‹ค์–‘ํ•œ ์• ํ”Œ๋ฆฌ์ผ€์ด์…˜์—์„œ YOLO11 ์„ ์‚ฌ์šฉํ•˜๊ธฐ ์œ„ํ•œ ์ˆ˜๋งŽ์€ ์˜ˆ์ œ์™€ ์‹ค์šฉ์ ์ธ ๊ฐ€์ด๋“œ๋ฅผ ์ œ๊ณตํ•ฉ๋‹ˆ๋‹ค. ์ข…ํ•ฉ์ ์ธ ๊ฐœ์š”๋ฅผ ๋ณด๋ ค๋ฉด Ultralytics ๋ธ”๋กœ๊ทธ์—์„œ ๊ฐ์ฒด ๊ฐ์ง€, ์„ธ๋ถ„ํ™” ๋“ฑ์„ ๋ณด์—ฌ์ฃผ๋Š” ์‚ฌ๋ก€ ์—ฐ๊ตฌ, ์ž์„ธํ•œ ์ž์Šต์„œ, ์ปค๋ฎค๋‹ˆํ‹ฐ ์Šคํ† ๋ฆฌ๋ฅผ YOLO11 ์—์„œ ํ™•์ธํ•˜์„ธ์š”. ๊ตฌ์ฒด์ ์ธ ์˜ˆ์‹œ๋Š” ์„ค๋ช…์„œ์˜ ์‚ฌ์šฉ๋ฒ• ์„น์…˜์„ ์ฐธ์กฐํ•˜์„ธ์š”.

๐Ÿ“…1 ๋…„ ์ „ ์ƒ์„ฑ๋จ โœ๏ธ 2๊ฐœ์›” ์ „ ์—…๋ฐ์ดํŠธ๋จ

๋Œ“๊ธ€