์ฝ˜ํ…์ธ ๋กœ ๊ฑด๋„ˆ๋›ฐ๊ธฐ

์ง€์›๋˜๋Š” ๋ชจ๋ธ Ultralytics

Ultralytics' ๋ชจ๋ธ ๋ฌธ์„œ์— ์˜ค์‹  ๊ฒƒ์„ ํ™˜์˜ํ•ฉ๋‹ˆ๋‹ค! ์ €ํฌ๋Š” ๋ฌผ์ฒด ๊ฐ์ง€, ์ธ์Šคํ„ด์Šค ๋ถ„ํ• , ์ด๋ฏธ์ง€ ๋ถ„๋ฅ˜, ํฌ์ฆˆ ์ถ”์ •, ๋‹ค์ค‘ ๋ฌผ์ฒด ์ถ”์ ๊ณผ ๊ฐ™์€ ํŠน์ • ์ž‘์—…์— ๋งž๊ฒŒ ์กฐ์ •๋œ ๋‹ค์–‘ํ•œ ๋ชจ๋ธ์„ ์ง€์›ํ•ฉ๋‹ˆ๋‹ค. ๋ชจ๋ธ ์•„ํ‚คํ…์ฒ˜๋ฅผ Ultralytics ์— ๊ธฐ์—ฌํ•˜๊ณ  ์‹ถ๋‹ค๋ฉด ๊ธฐ์—ฌ ๊ฐ€์ด๋“œ๋ฅผ ํ™•์ธํ•˜์„ธ์š”.

๋‹ค์Œ์€ ์ง€์›๋˜๋Š” ์ฃผ์š” ๋ชจ๋ธ ์ค‘ ์ผ๋ถ€์ž…๋‹ˆ๋‹ค:

  1. YOLOv3: ํšจ์œจ์ ์ธ ์‹ค์‹œ๊ฐ„ ๊ฐ์ฒด ๊ฐ์ง€ ๊ธฐ๋Šฅ์œผ๋กœ ์œ ๋ช…ํ•œ ์กฐ์…‰ ๋ ˆ๋“œ๋ชฌ์ด ๊ฐœ๋ฐœํ•œ YOLO ๋ชจ๋ธ ์ œํ’ˆ๊ตฐ์˜ ์„ธ ๋ฒˆ์งธ ๋ฐ˜๋ณต ๋ฒ„์ „์ž…๋‹ˆ๋‹ค.
  2. YOLOv4: ์•Œ๋ ‰์„ธ์ด ๋ณดํ์ฝ”๋ธŒ์Šคํ‚ค๊ฐ€ 2020๋…„์— ์ถœ์‹œํ•œ YOLOv3์˜ ๋‹คํฌ๋„ท ๋„ค์ดํ‹ฐ๋ธŒ ์—…๋ฐ์ดํŠธ์ž…๋‹ˆ๋‹ค.
  3. YOLOv5: Ultralytics ์˜ YOLO ์•„ํ‚คํ…์ฒ˜ ๊ฐœ์„  ๋ฒ„์ „์œผ๋กœ, ์ด์ „ ๋ฒ„์ „์— ๋น„ํ•ด ์„ฑ๋Šฅ๊ณผ ์†๋„ ๋ฉด์—์„œ ๋” ๋‚˜์€ ์ ˆ์ถฉ์•ˆ์„ ์ œ๊ณตํ•ฉ๋‹ˆ๋‹ค.
  4. YOLOv6: ๋ฉ”์ดํ‡€์ด 2022๋…„์— ์ถœ์‹œํ–ˆ์œผ๋ฉฐ, ๋ฉ”์ดํ‡€์˜ ์—ฌ๋Ÿฌ ์ž์œจ์ฃผํ–‰ ๋ฐฐ์†ก ๋กœ๋ด‡์— ์‚ฌ์šฉ๋˜๊ณ  ์žˆ์Šต๋‹ˆ๋‹ค.
  5. YOLOv7: ์—…๋ฐ์ดํŠธ๋จ YOLO YOLOv4์˜ ์ €์ž๊ฐ€ 2022๋…„์— ์ถœ์‹œํ•œ ๋ชจ๋ธ์ž…๋‹ˆ๋‹ค.
  6. YOLOv8 NEW ๐Ÿš€: ์ธ์Šคํ„ด์Šค ์„ธ๋ถ„ํ™”, ํฌ์ฆˆ/ํ‚คํฌ์ธํŠธ ์ถ”์ •, ๋ถ„๋ฅ˜ ๋“ฑ ํ–ฅ์ƒ๋œ ๊ธฐ๋Šฅ์„ ๊ฐ–์ถ˜ YOLO ์ œํ’ˆ๊ตฐ์˜ ์ตœ์‹  ๋ฒ„์ „์ž…๋‹ˆ๋‹ค.
  7. YOLOv9: ํ”„๋กœ๊ทธ๋ž˜๋ฐ ๊ฐ€๋Šฅํ•œ ๊ธฐ์šธ๊ธฐ ์ •๋ณด๋ฅผ ๊ตฌํ˜„ํ•˜๋Š” Ultralytics YOLOv5 ํ”„๋กœ๊ทธ๋ž˜๋จธ๋ธ” ๊ทธ๋ผ๋ฐ์ด์…˜ ์ •๋ณด(PGI)๋ฅผ ๊ตฌํ˜„ํ•˜๋Š” ์ฝ”๋“œ๋ฒ ์ด์Šค.
  8. ๋ฌด์—‡์ด๋“  ์„ธ๊ทธ๋จผํŠธ ๋ชจ๋ธ (SAM): ๋ฉ”ํƒ€์˜ ์„ธ๊ทธ๋จผํŠธ ์• ๋‹ˆ์”ฝ ๋ชจ๋ธ (SAM).
  9. ๋ชจ๋ฐ”์ผ ์„ธ๊ทธ๋จผํŠธ ์• ๋‹ˆ์”ฝ ๋ชจ๋ธ (MobileSAM)๊ฒฝํฌ๋Œ€ํ•™๊ต ๋ชจ๋ฐ”์ผ ์• ํ”Œ๋ฆฌ์ผ€์ด์…˜ : MobileSAM .
  10. ๋น ๋ฅธ ์„ธ๊ทธ๋จผํŠธ ๋ฌด์—‡์ด๋“  ๋ชจ๋ธ (FastSAM)์ค‘๊ตญ๊ณผํ•™์› ์ž๋™ํ™”์—ฐ๊ตฌ์†Œ ์ด๋ฏธ์ง€ ๋ฐ ๋น„๋””์˜ค ๋ถ„์„ ๊ทธ๋ฃน( FastSAM ).
  11. YOLO-NAS: YOLO ์‹ ๊ฒฝ๋ง ์•„ํ‚คํ…์ฒ˜ ๊ฒ€์ƒ‰(NAS) ๋ชจ๋ธ.
  12. ์‹ค์‹œ๊ฐ„ ๊ฐ์ง€ ํŠธ๋žœ์Šคํฌ๋จธ (RT-DETR): ๋ฐ”์ด๋‘์˜ PaddlePaddle ์‹ค์‹œ๊ฐ„ ๊ฐ์ง€ ํŠธ๋žœ์Šคํฌ๋จธ (RT-DETR) ๋ชจ๋ธ.
  13. YOLO-์„ธ๊ณ„: ์‹ค์‹œ๊ฐ„ ๊ฐœ๋ฐฉํ˜• ์–ดํœ˜ ๊ฐœ์ฒด ๊ฐ์ง€ ๋ชจ๋ธ: Tencent AI Lab.



Watch: ๋ช‡ ์ค„์˜ ์ฝ”๋“œ๋งŒ์œผ๋กœ Ultralytics YOLO ๋ชจ๋ธ์„ ์‹คํ–‰ํ•˜์„ธ์š”.

์‹œ์ž‘ํ•˜๊ธฐ: ์‚ฌ์šฉ ์˜ˆ์‹œ

์ด ์˜ˆ๋Š” ๊ฐ„๋‹จํ•œ YOLO ํ›ˆ๋ จ ๋ฐ ์ถ”๋ก  ์˜ˆ์ œ๋ฅผ ์ œ๊ณตํ•ฉ๋‹ˆ๋‹ค. ์ด๋Ÿฌํ•œ ๋ชจ๋“œ ๋ฐ ๊ธฐํƒ€ ๋ชจ๋“œ์— ๋Œ€ํ•œ ์ „์ฒด ์„ค๋ช…์„œ๋Š” ์˜ˆ์ธก, ํ•™์Šต, Val ๋ฐ ๋‚ด๋ณด๋‚ด๊ธฐ ๋ฌธ์„œ ํŽ˜์ด์ง€๋ฅผ ์ฐธ์กฐํ•˜์„ธ์š”.

์•„๋ž˜ ์˜ˆ๋Š” YOLOv8 ๊ฐ์ฒด ๊ฐ์ง€๋ฅผ ์œ„ํ•œ ๋ชจ๋ธ๊ฐ์ง€์šฉ ์˜ˆ์ œ์ž…๋‹ˆ๋‹ค. ์ง€์›๋˜๋Š” ์ถ”๊ฐ€ ์ž‘์—…์— ๋Œ€ํ•ด์„œ๋Š” ์„ธ๊ทธ๋จผํŠธ, ๋ถ„๋ฅ˜ ๋ฐ ํฌ์ฆˆ ๋ฌธ์„œ๋ฅผ ์ฐธ์กฐํ•˜์„ธ์š”.

์˜ˆ์ œ

PyTorch ์‚ฌ์ „ ๊ต์œก *.pt ๋ชจ๋ธ ๋ฐ ๊ตฌ์„ฑ *.yaml ํŒŒ์ผ์— ์ „๋‹ฌํ•  ์ˆ˜ ์žˆ์Šต๋‹ˆ๋‹ค. YOLO(), SAM(), NAS() ๋ฐ RTDETR() ํด๋ž˜์Šค๋ฅผ ์‚ฌ์šฉํ•˜์—ฌ Python ์—์„œ ๋ชจ๋ธ ์ธ์Šคํ„ด์Šค๋ฅผ ์ƒ์„ฑํ•ฉ๋‹ˆ๋‹ค:

from ultralytics import YOLO

# Load a COCO-pretrained YOLOv8n model
model = YOLO('yolov8n.pt')

# Display model information (optional)
model.info()

# Train the model on the COCO8 example dataset for 100 epochs
results = model.train(data='coco8.yaml', epochs=100, imgsz=640)

# Run inference with the YOLOv8n model on the 'bus.jpg' image
results = model('path/to/bus.jpg')

CLI ๋ช…๋ น์„ ์‚ฌ์šฉํ•˜์—ฌ ๋ชจ๋ธ์„ ์ง์ ‘ ์‹คํ–‰ํ•  ์ˆ˜ ์žˆ์Šต๋‹ˆ๋‹ค:

# Load a COCO-pretrained YOLOv8n model and train it on the COCO8 example dataset for 100 epochs
yolo train model=yolov8n.pt data=coco8.yaml epochs=100 imgsz=640

# Load a COCO-pretrained YOLOv8n model and run inference on the 'bus.jpg' image
yolo predict model=yolov8n.pt source=path/to/bus.jpg

์ƒˆ ๋ชจ๋ธ ๊ธฐ์—ฌํ•˜๊ธฐ

Ultralytics ์— ๋ชจ๋ธ ๊ธฐ๊ณ ์— ๊ด€์‹ฌ์ด ์žˆ์œผ์‹ ๊ฐ€์š”? ์ข‹์•„์š”! ์ €ํฌ๋Š” ํ•ญ์ƒ ๋ชจ๋ธ ํฌํŠธํด๋ฆฌ์˜ค๋ฅผ ํ™•์žฅํ•  ์ˆ˜ ์žˆ๋Š” ๊ธฐํšŒ๋ฅผ ์—ด์–ด๋‘๊ณ  ์žˆ์Šต๋‹ˆ๋‹ค.

  1. ๋ฆฌํฌ์ง€ํ† ๋ฆฌ ํฌํฌ: Ultralytics GitHub ๋ฆฌํฌ์ง€ํ† ๋ฆฌ๋ฅผ ํฌํฌํ•˜์—ฌ ์‹œ์ž‘ํ•˜์„ธ์š”.

  2. ํฌํฌ ๋ณต์ œ: ํฌํฌ๋ฅผ ๋กœ์ปฌ ๋จธ์‹ ์— ๋ณต์ œํ•˜๊ณ  ์ž‘์—…ํ•  ์ƒˆ ๋ธŒ๋žœ์น˜๋ฅผ ๋งŒ๋“ญ๋‹ˆ๋‹ค.

  3. ๋ชจ๋ธ ๊ตฌํ˜„ํ•˜๊ธฐ: ๊ธฐ์—ฌ ๊ฐ€์ด๋“œ์— ์ œ๊ณต๋œ ์ฝ”๋”ฉ ํ‘œ์ค€๊ณผ ๊ฐ€์ด๋“œ๋ผ์ธ์— ๋”ฐ๋ผ ๋ชจ๋ธ์„ ์ถ”๊ฐ€ํ•˜์„ธ์š”.

  4. ์ฒ ์ €ํ•œ ํ…Œ์ŠคํŠธ: ๋ชจ๋ธ์„ ๊ฐœ๋ณ„์ ์œผ๋กœ ๋˜๋Š” ํŒŒ์ดํ”„๋ผ์ธ์˜ ์ผ๋ถ€๋กœ ์—„๊ฒฉํ•˜๊ฒŒ ํ…Œ์ŠคํŠธํ•ด์•ผ ํ•ฉ๋‹ˆ๋‹ค.

  5. ํ’€ ๋ฆฌํ€˜์ŠคํŠธ๋ฅผ ์ƒ์„ฑํ•ฉ๋‹ˆ๋‹ค: ๋ชจ๋ธ์— ๋งŒ์กฑํ•˜๋ฉด ๋ฉ”์ธ ๋ฆฌํฌ์ง€ํ† ๋ฆฌ์— ํ’€ ๋ฆฌํ€˜์ŠคํŠธ๋ฅผ ๋งŒ๋“ค์–ด ๊ฒ€ํ† ๋ฅผ ์š”์ฒญํ•˜์„ธ์š”.

  6. ์ฝ”๋“œ ๊ฒ€ํ†  ๋ฐ ๋ณ‘ํ•ฉ: ๊ฒ€ํ†  ํ›„ ๋ชจ๋ธ์ด ๊ธฐ์ค€์„ ์ถฉ์กฑํ•˜๋ฉด ๋ฉ”์ธ ๋ฆฌํฌ์ง€ํ† ๋ฆฌ์— ๋ณ‘ํ•ฉ๋ฉ๋‹ˆ๋‹ค.

์ž์„ธํ•œ ๋‹จ๊ณ„๋Š” ๊ธฐ์—ฌ ๊ฐ€์ด๋“œ๋ฅผ ์ฐธ์กฐํ•˜์„ธ์š”.



์ƒ์„ฑ 2023-11-12, ์—…๋ฐ์ดํŠธ 2024-02-26
์ž‘์„ฑ์ž: glenn-jocher (7), Laughing-q (1)

๋Œ“๊ธ€