์ฝ˜ํ…์ธ ๋กœ ๊ฑด๋„ˆ๋›ฐ๊ธฐ

ํ˜ธ๋ž‘์ด ํฌ์ฆˆ ๋ฐ์ดํ„ฐ ์„ธํŠธ

์†Œ๊ฐœ

Ultralytics ์—์„œ๋Š” ํฌ์ฆˆ ์ถ”์ • ์ž‘์—…์„ ์œ„ํ•ด ์„ค๊ณ„๋œ ๋‹ค์šฉ๋„ ์ปฌ๋ ‰์…˜์ธ Tiger-Pose ๋ฐ์ดํ„ฐ ์„ธํŠธ๋ฅผ ์†Œ๊ฐœํ•ฉ๋‹ˆ๋‹ค. ์ด ๋ฐ์ดํ„ฐ ์„ธํŠธ๋Š” YouTube ๋™์˜์ƒ์—์„œ ๊ฐ€์ ธ์˜จ 263๊ฐœ์˜ ์ด๋ฏธ์ง€๋กœ ๊ตฌ์„ฑ๋˜์–ด ์žˆ์œผ๋ฉฐ, 210๊ฐœ์˜ ์ด๋ฏธ์ง€๋Š” ํ›ˆ๋ จ์šฉ์œผ๋กœ, 53๊ฐœ๋Š” ๊ฒ€์ฆ์šฉ์œผ๋กœ ํ• ๋‹น๋˜์–ด ์žˆ์Šต๋‹ˆ๋‹ค. ํฌ์ฆˆ ์ถ”์ • ์•Œ๊ณ ๋ฆฌ์ฆ˜์„ ํ…Œ์ŠคํŠธํ•˜๊ณ  ๋ฌธ์ œ๋ฅผ ํ•ด๊ฒฐํ•˜๊ธฐ ์œ„ํ•œ ํ›Œ๋ฅญํ•œ ๋ฆฌ์†Œ์Šค์ž…๋‹ˆ๋‹ค.

ํ˜ธ๋ž‘์ด ํฌ์ฆˆ ๋ฐ์ดํ„ฐ ์„ธํŠธ๋Š” 210๊ฐœ์˜ ์ด๋ฏธ์ง€๋กœ ๊ด€๋ฆฌํ•˜๊ธฐ ์‰ฌ์šด ํฌ๊ธฐ์ž„์—๋„ ๋ถˆ๊ตฌํ•˜๊ณ  ๋‹ค์–‘์„ฑ์„ ์ œ๊ณตํ•˜๊ธฐ ๋•Œ๋ฌธ์— ํ›ˆ๋ จ ํŒŒ์ดํ”„๋ผ์ธ์„ ํ‰๊ฐ€ํ•˜๊ณ  ์ž ์žฌ์ ์ธ ์˜ค๋ฅ˜๋ฅผ ์‹๋ณ„ํ•˜๋ฉฐ ํฌ์ฆˆ ์ถ”์ •์„ ์œ„ํ•ด ๋” ํฐ ๋ฐ์ดํ„ฐ ์„ธํŠธ๋กœ ์ž‘์—…ํ•˜๊ธฐ ์ „์— ์ค‘์š”ํ•œ ์˜ˆ๋น„ ๋‹จ๊ณ„๋กœ ์‚ฌ์šฉํ•˜๊ธฐ์— ์ ํ•ฉํ•ฉ๋‹ˆ๋‹ค.

์ด ๋ฐ์ดํ„ฐ ์„ธํŠธ๋Š” ๋‹ค์Œ๊ณผ ํ•จ๊ป˜ ์‚ฌ์šฉํ•˜๊ธฐ ์œ„ํ•œ ๊ฒƒ์ž…๋‹ˆ๋‹ค. Ultralytics HUB ๋ฐ YOLOv8.



Watch: ๋‹ค์Œ์„ ์‚ฌ์šฉํ•˜์—ฌ ํƒ€์ด๊ฑฐ ํฌ์ฆˆ ๋ฐ์ดํ„ฐ ์„ธํŠธ์—์„œ YOLOv8 ํฌ์ฆˆ ๋ชจ๋ธ ํ›ˆ๋ จํ•˜๊ธฐ Ultralytics HUB

๋ฐ์ดํ„ฐ ์ง‘ํ•ฉ YAML

YAML(๋˜ ๋‹ค๋ฅธ ๋งˆํฌ์—… ์–ธ์–ด) ํŒŒ์ผ์€ ๋ฐ์ดํ„ฐ ์„ธํŠธ์˜ ๊ตฌ์„ฑ ์„ธ๋ถ€ ์ •๋ณด๋ฅผ ์ง€์ •ํ•˜๋Š” ์ˆ˜๋‹จ์œผ๋กœ ์‚ฌ์šฉ๋ฉ๋‹ˆ๋‹ค. ์—ฌ๊ธฐ์—๋Š” ํŒŒ์ผ ๊ฒฝ๋กœ, ํด๋ž˜์Šค ์ •์˜ ๋ฐ ๊ธฐํƒ€ ๊ด€๋ จ ์ •๋ณด์™€ ๊ฐ™์€ ์ค‘์š”ํ•œ ๋ฐ์ดํ„ฐ๊ฐ€ ํฌํ•จ๋ฉ๋‹ˆ๋‹ค. ํŠนํžˆ tiger-pose.yaml ํŒŒ์ผ์—์„œ ๋‹ค์Œ์„ ํ™•์ธํ•  ์ˆ˜ ์žˆ์Šต๋‹ˆ๋‹ค. Ultralytics ํƒ€์ด๊ฑฐ ํฌ์ฆˆ ๋ฐ์ดํ„ฐ ์„ธํŠธ ๊ตฌ์„ฑ ํŒŒ์ผ.

ultralytics/cfg/๋ฐ์ดํ„ฐ์„ธํŠธ/tiger-pose.yaml

# Ultralytics YOLO ๐Ÿš€, AGPL-3.0 license
# Tiger Pose dataset by Ultralytics
# Documentation: https://docs.ultralytics.com/datasets/pose/tiger-pose/
# Example usage: yolo train data=tiger-pose.yaml
# parent
# โ”œโ”€โ”€ ultralytics
# โ””โ”€โ”€ datasets
#     โ””โ”€โ”€ tiger-pose  โ† downloads here (75.3 MB)

# 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/tiger-pose # dataset root dir
train: train # train images (relative to 'path') 210 images
val: val # val images (relative to 'path') 53 images

# Keypoints
kpt_shape: [12, 2] # number of keypoints, number of dims (2 for x,y or 3 for x,y,visible)
flip_idx: [0, 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11]

# Classes
names:
  0: tiger

# Download script/URL (optional)
download: https://ultralytics.com/assets/tiger-pose.zip

์‚ฌ์šฉ๋ฒ•

์ด๋ฏธ์ง€ ํฌ๊ธฐ๊ฐ€ 640์ธ ํƒ€์ด๊ฑฐ ํฌ์ฆˆ ๋ฐ์ดํ„ฐ ์„ธํŠธ์— ๋Œ€ํ•ด 100๊ฐœ์˜ ์—ํฌํฌ์— ๋Œ€ํ•ด YOLOv8n-pose ๋ชจ๋ธ์„ ํ›ˆ๋ จํ•˜๋ ค๋ฉด ๋‹ค์Œ ์ฝ”๋“œ ์กฐ๊ฐ์„ ์‚ฌ์šฉํ•  ์ˆ˜ ์žˆ์Šต๋‹ˆ๋‹ค. ์‚ฌ์šฉ ๊ฐ€๋Šฅํ•œ ์ธ์ˆ˜์˜ ์ „์ฒด ๋ชฉ๋ก์€ ๋ชจ๋ธ ํ›ˆ๋ จ ํŽ˜์ด์ง€๋ฅผ ์ฐธ์กฐํ•˜์„ธ์š”.

์—ด์ฐจ ์˜ˆ์‹œ

from ultralytics import YOLO

# Load a model
model = YOLO('yolov8n-pose.pt')  # load a pretrained model (recommended for training)

# Train the model
results = model.train(data='tiger-pose.yaml', epochs=100, imgsz=640)
# Start training from a pretrained *.pt model
yolo task=pose mode=train data=tiger-pose.yaml model=yolov8n.pt epochs=100 imgsz=640

์ƒ˜ํ”Œ ์ด๋ฏธ์ง€ ๋ฐ ์ฃผ์„

๋‹ค์Œ์€ ํ˜ธ๋ž‘์ด ํฌ์ฆˆ ๋ฐ์ดํ„ฐ ์„ธํŠธ์˜ ์ด๋ฏธ์ง€์™€ ํ•ด๋‹น ์ฃผ์„์˜ ๋ช‡ ๊ฐ€์ง€ ์˜ˆ์ž…๋‹ˆ๋‹ค:

๋ฐ์ดํ„ฐ ์„ธํŠธ ์ƒ˜ํ”Œ ์ด๋ฏธ์ง€

  • ๋ชจ์ž์ดํฌ ์ด๋ฏธ์ง€: ์ด ์ด๋ฏธ์ง€๋Š” ๋ชจ์ž์ดํฌ๋œ ๋ฐ์ดํ„ฐ ์„ธํŠธ ์ด๋ฏธ์ง€๋กœ ๊ตฌ์„ฑ๋œ ํ›ˆ๋ จ ๋ฐฐ์น˜๋ฅผ ๋ณด์—ฌ์ค๋‹ˆ๋‹ค. ๋ชจ์ž์ดํฌ๋Š” ์—ฌ๋Ÿฌ ์ด๋ฏธ์ง€๋ฅผ ๋‹จ์ผ ์ด๋ฏธ์ง€๋กœ ๊ฒฐํ•ฉํ•˜์—ฌ ๊ฐ ํ›ˆ๋ จ ๋ฐฐ์น˜ ๋‚ด์—์„œ ๋‹ค์–‘ํ•œ ๊ฐœ์ฒด์™€ ์žฅ๋ฉด์„ ๋Š˜๋ฆฌ๋Š” ํ›ˆ๋ จ ์ค‘์— ์‚ฌ์šฉ๋˜๋Š” ๊ธฐ๋ฒ•์ž…๋‹ˆ๋‹ค. ์ด๋ฅผ ํ†ตํ•ด ๋‹ค์–‘ํ•œ ๊ฐ์ฒด ํฌ๊ธฐ, ์ข…ํšก๋น„ ๋ฐ ์ปจํ…์ŠคํŠธ์— ์ผ๋ฐ˜ํ™”ํ•  ์ˆ˜ ์žˆ๋Š” ๋ชจ๋ธ์˜ ๋Šฅ๋ ฅ์„ ํ–ฅ์ƒ์‹œํ‚ฌ ์ˆ˜ ์žˆ์Šต๋‹ˆ๋‹ค.

์ด ์˜ˆ์‹œ๋Š” ํƒ€์ด๊ฑฐ ํฌ์ฆˆ ๋ฐ์ดํ„ฐ ์„ธํŠธ์— ํฌํ•จ๋œ ์ด๋ฏธ์ง€์˜ ๋‹ค์–‘์„ฑ๊ณผ ๋ณต์žก์„ฑ, ๊ทธ๋ฆฌ๊ณ  ํ›ˆ๋ จ ๊ณผ์ •์—์„œ ๋ชจ์ž์ดํฌ ์ฒ˜๋ฆฌ์˜ ์ด์ ์„ ๋ณด์—ฌ์ค๋‹ˆ๋‹ค.

์ถ”๋ก  ์˜ˆ์ œ

์ถ”๋ก  ์˜ˆ์ œ

from ultralytics import YOLO

# Load a model
model = YOLO('path/to/best.pt')  # load a tiger-pose trained model

# Run inference
results = model.predict(source="https://www.youtube.com/watch?v=MIBAT6BGE6U&pp=ygUYdGlnZXIgd2Fsa2luZyByZWZlcmVuY2Ug" show=True)
# Run inference using a tiger-pose trained model
yolo task=pose mode=predict source="https://www.youtube.com/watch?v=MIBAT6BGE6U&pp=ygUYdGlnZXIgd2Fsa2luZyByZWZlcmVuY2Ug" show=True model="path/to/best.pt"

์ธ์šฉ ๋ฐ ๊ฐ์‚ฌ

์ด ๋ฐ์ดํ„ฐ ์„ธํŠธ๋Š” AGPL-3.0 ๋ผ์ด์„ ์Šค์— ๋”ฐ๋ผ ๊ณต๊ฐœ๋˜์—ˆ์Šต๋‹ˆ๋‹ค.



์ƒ์„ฑ๋จ 2023-11-12, ์—…๋ฐ์ดํŠธ๋จ 2024-02-03
์ž‘์„ฑ์ž: glenn-jocher (5), RizwanMunawar (1)

๋Œ“๊ธ€