์ฝ˜ํ…์ธ ๋กœ ๊ฑด๋„ˆ๋›ฐ๊ธฐ

MNIST ๋ฐ์ดํ„ฐ ์„ธํŠธ

MNIST (๋ฏธ๊ตญ ๊ตญ๋ฆฝํ‘œ์ค€๊ธฐ์ˆ ์—ฐ๊ตฌ์†Œ) ๋ฐ์ดํ„ฐ ์„ธํŠธ๋Š” ๋‹ค์–‘ํ•œ ์ด๋ฏธ์ง€ ์ฒ˜๋ฆฌ ์‹œ์Šคํ…œ๊ณผ ๋จธ์‹  ๋Ÿฌ๋‹ ๋ชจ๋ธ์„ ํ›ˆ๋ จํ•˜๋Š” ๋ฐ ์ผ๋ฐ˜์ ์œผ๋กœ ์‚ฌ์šฉ๋˜๋Š” ๋Œ€๊ทœ๋ชจ ํ•„๊ธฐ ์ˆซ์ž ๋ฐ์ดํ„ฐ๋ฒ ์ด์Šค์ž…๋‹ˆ๋‹ค. NIST์˜ ์›๋ณธ ๋ฐ์ดํ„ฐ ์„ธํŠธ์—์„œ ์ƒ˜ํ”Œ์„ "์žฌํ˜ผํ•ฉ"ํ•˜์—ฌ ์ƒ์„ฑ๋˜์—ˆ์œผ๋ฉฐ ์ด๋ฏธ์ง€ ๋ถ„๋ฅ˜ ์•Œ๊ณ ๋ฆฌ์ฆ˜์˜ ์„ฑ๋Šฅ์„ ํ‰๊ฐ€ํ•˜๋Š” ๋ฒค์น˜๋งˆํฌ๊ฐ€ ๋˜์—ˆ์Šต๋‹ˆ๋‹ค.

์ฃผ์š” ๊ธฐ๋Šฅ

  • MNIST์—๋Š” 60,000๊ฐœ์˜ ํ›ˆ๋ จ ์ด๋ฏธ์ง€์™€ 10,000๊ฐœ์˜ ์†๊ธ€์”จ ์ˆซ์ž ํ…Œ์ŠคํŠธ ์ด๋ฏธ์ง€๊ฐ€ ํฌํ•จ๋˜์–ด ์žˆ์Šต๋‹ˆ๋‹ค.
  • ๋ฐ์ดํ„ฐ ์„ธํŠธ๋Š” 28x28 ํ”ฝ์…€ ํฌ๊ธฐ์˜ ๊ทธ๋ ˆ์ด ์Šค์ผ€์ผ ์ด๋ฏธ์ง€๋กœ ๊ตฌ์„ฑ๋ฉ๋‹ˆ๋‹ค.
  • The images are normalized to fit into a 28x28 pixel bounding box and anti-aliased, introducing grayscale levels.
  • MNIST๋Š” ๋จธ์‹ ๋Ÿฌ๋‹ ๋ถ„์•ผ, ํŠนํžˆ ์ด๋ฏธ์ง€ ๋ถ„๋ฅ˜ ์ž‘์—…์˜ ํ›ˆ๋ จ ๋ฐ ํ…Œ์ŠคํŠธ์— ๋„๋ฆฌ ์‚ฌ์šฉ๋ฉ๋‹ˆ๋‹ค.

๋ฐ์ดํ„ฐ ์„ธํŠธ ๊ตฌ์กฐ

MNIST ๋ฐ์ดํ„ฐ ์„ธํŠธ๋Š” ๋‘ ๊ฐœ์˜ ํ•˜์œ„ ์ง‘ํ•ฉ์œผ๋กœ ๋‚˜๋‰ฉ๋‹ˆ๋‹ค:

  1. ํ›ˆ๋ จ ์ง‘ํ•ฉ: ์ด ํ•˜์œ„ ์ง‘ํ•ฉ์—๋Š” ๋จธ์‹  ๋Ÿฌ๋‹ ๋ชจ๋ธ ํ•™์Šต์— ์‚ฌ์šฉ๋˜๋Š” 60,000๊ฐœ์˜ ์†๊ธ€์”จ ์ˆซ์ž ์ด๋ฏธ์ง€๊ฐ€ ํฌํ•จ๋˜์–ด ์žˆ์Šต๋‹ˆ๋‹ค.
  2. ํ…Œ์ŠคํŠธ ์„ธํŠธ: ์ด ํ•˜์œ„ ์ง‘ํ•ฉ์€ ํ•™์Šต๋œ ๋ชจ๋ธ์„ ํ…Œ์ŠคํŠธํ•˜๊ณ  ๋ฒค์น˜๋งˆํ‚นํ•˜๋Š” ๋ฐ ์‚ฌ์šฉ๋˜๋Š” 10,000๊ฐœ์˜ ์ด๋ฏธ์ง€๋กœ ๊ตฌ์„ฑ๋ฉ๋‹ˆ๋‹ค.

ํ™•์žฅ MNIST(EMNIST)

ํ™•์žฅ๋œ MNIST(EMNIST)๋Š” MNIST์˜ ํ›„์†์œผ๋กœ NIST์—์„œ ๊ฐœ๋ฐœํ•˜์—ฌ ๊ณต๊ฐœํ•œ ์ตœ์‹  ๋ฐ์ดํ„ฐ ์„ธํŠธ์ž…๋‹ˆ๋‹ค. MNIST์—๋Š” ์†์œผ๋กœ ์“ด ์ˆซ์ž์˜ ์ด๋ฏธ์ง€๋งŒ ํฌํ•จ๋˜์–ด ์žˆ๋Š” ๋ฐ˜๋ฉด, EMNIST์—๋Š” ์ˆซ์ž๋Š” ๋ฌผ๋ก  ์†์œผ๋กœ ์“ด ๋Œ€๋ฌธ์ž์™€ ์†Œ๋ฌธ์ž์˜ ๋Œ€๊ทœ๋ชจ ๋ฐ์ดํ„ฐ๋ฒ ์ด์Šค์ธ NIST ํŠน์ˆ˜ ๋ฐ์ดํ„ฐ๋ฒ ์ด์Šค 19์˜ ๋ชจ๋“  ์ด๋ฏธ์ง€๊ฐ€ ํฌํ•จ๋˜์–ด ์žˆ์Šต๋‹ˆ๋‹ค. EMNIST์˜ ์ด๋ฏธ์ง€๋Š” MNIST ์ด๋ฏธ์ง€์™€ ๋™์ผํ•œ ํ”„๋กœ์„ธ์Šค๋ฅผ ํ†ตํ•ด ๋™์ผํ•œ 28x28ํ”ฝ์…€ ํฌ๋งท์œผ๋กœ ๋ณ€ํ™˜๋˜์—ˆ์Šต๋‹ˆ๋‹ค. ๋”ฐ๋ผ์„œ ์ด์ „์˜ ๋” ์ž‘์€ MNIST ๋ฐ์ดํ„ฐ ์„ธํŠธ์—์„œ ์ž‘๋™ํ•˜๋Š” ๋„๊ตฌ๋Š” EMNIST์—์„œ๋„ ์ˆ˜์ •ํ•˜์ง€ ์•Š๊ณ  ์ž‘๋™ํ•  ๊ฐ€๋Šฅ์„ฑ์ด ๋†’์Šต๋‹ˆ๋‹ค.

์• ํ”Œ๋ฆฌ์ผ€์ด์…˜

The MNIST dataset is widely used for training and evaluating deep learning models in image classification tasks, such as Convolutional Neural Networks (CNNs), Support Vector Machines (SVMs), and various other machine learning algorithms. The dataset's simple and well-structured format makes it an essential resource for researchers and practitioners in the field of machine learning and computer vision.

์‚ฌ์šฉ๋ฒ•

To train a CNN model on the MNIST dataset for 100 epochs with an image size of 32x32, you can use the following code snippets. For a comprehensive list of available arguments, refer to the model Training page.

์—ด์ฐจ ์˜ˆ์‹œ

from ultralytics import YOLO

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

# Train the model
results = model.train(data="mnist", epochs=100, imgsz=32)
# Start training from a pretrained *.pt model
yolo classify train data=mnist model=yolo11n-cls.pt epochs=100 imgsz=28

์ƒ˜ํ”Œ ์ด๋ฏธ์ง€ ๋ฐ ์ฃผ์„

The MNIST dataset contains grayscale images of handwritten digits, providing a well-structured dataset for image classification tasks. Here are some examples of images from the dataset:

๋ฐ์ดํ„ฐ ์„ธํŠธ ์ƒ˜ํ”Œ ์ด๋ฏธ์ง€

์ด ์˜ˆ๋Š” MNIST ๋ฐ์ดํ„ฐ ์„ธํŠธ์— ํฌํ•จ๋œ ํ•„๊ธฐ ์ˆซ์ž์˜ ๋‹ค์–‘์„ฑ๊ณผ ๋ณต์žก์„ฑ์„ ๋ณด์—ฌ์ฃผ๋ฉฐ, ๊ฐ•๋ ฅํ•œ ์ด๋ฏธ์ง€ ๋ถ„๋ฅ˜ ๋ชจ๋ธ์„ ํ›ˆ๋ จํ•˜๊ธฐ ์œ„ํ•œ ๋‹ค์–‘ํ•œ ๋ฐ์ดํ„ฐ ์„ธํŠธ์˜ ์ค‘์š”์„ฑ์„ ๊ฐ•์กฐํ•ฉ๋‹ˆ๋‹ค.

์ธ์šฉ ๋ฐ ๊ฐ์‚ฌ

MNIST ๋ฐ์ดํ„ฐ ์ง‘ํ•ฉ์„ ์‚ฌ์šฉํ•˜๋Š” ๊ฒฝ์šฐ

์—ฐ๊ตฌ ๋˜๋Š” ๊ฐœ๋ฐœ ์ž‘์—…์˜ ๊ฒฝ์šฐ ๋‹ค์Œ ๋…ผ๋ฌธ์„ ์ธ์šฉํ•ด ์ฃผ์„ธ์š”:

@article{lecun2010mnist,
         title={MNIST handwritten digit database},
         author={LeCun, Yann and Cortes, Corinna and Burges, CJ},
         journal={ATT Labs [Online]. Available: http://yann.lecun.com/exdb/mnist},
         volume={2},
         year={2010}
}

We would like to acknowledge Yann LeCun, Corinna Cortes, and Christopher J.C. Burges for creating and maintaining the MNIST dataset as a valuable resource for the machine learning and computer vision research community. For more information about the MNIST dataset and its creators, visit the MNIST dataset website.

์ž์ฃผ ๋ฌป๋Š” ์งˆ๋ฌธ

MNIST ๋ฐ์ดํ„ฐ ์„ธํŠธ๋Š” ๋ฌด์—‡์ด๋ฉฐ ๋จธ์‹  ๋Ÿฌ๋‹์—์„œ ์ค‘์š”ํ•œ ์ด์œ ๋Š” ๋ฌด์—‡์ธ๊ฐ€์š”?

MNIST ๋ฐ์ดํ„ฐ ์„ธํŠธ ๋˜๋Š” ์ˆ˜์ •๋œ ๊ตญ๋ฆฝํ‘œ์ค€๊ธฐ์ˆ ์—ฐ๊ตฌ์†Œ ๋ฐ์ดํ„ฐ ์„ธํŠธ๋Š” ์ด๋ฏธ์ง€ ๋ถ„๋ฅ˜ ์‹œ์Šคํ…œ์˜ ํ›ˆ๋ จ ๋ฐ ํ…Œ์ŠคํŠธ๋ฅผ ์œ„ํ•ด ์„ค๊ณ„๋œ ๋„๋ฆฌ ์‚ฌ์šฉ๋˜๋Š” ์ˆ˜๊ธฐ ์ˆซ์ž ๋ชจ์Œ์ž…๋‹ˆ๋‹ค. ์—ฌ๊ธฐ์—๋Š” 60,000๊ฐœ์˜ ํ›ˆ๋ จ ์ด๋ฏธ์ง€์™€ 10,000๊ฐœ์˜ ํ…Œ์ŠคํŠธ ์ด๋ฏธ์ง€๊ฐ€ ํฌํ•จ๋˜์–ด ์žˆ์œผ๋ฉฐ, ๋ชจ๋‘ 28x28ํ”ฝ์…€ ํฌ๊ธฐ์˜ ํšŒ์ƒ‰์กฐ ์ด๋ฏธ์ง€์ž…๋‹ˆ๋‹ค. ์ด ๋ฐ์ดํ„ฐ ์„ธํŠธ์˜ ์ค‘์š”์„ฑ์€ ์ด๋ฏธ์ง€ ๋ถ„๋ฅ˜ ์•Œ๊ณ ๋ฆฌ์ฆ˜์„ ํ‰๊ฐ€ํ•˜๊ธฐ ์œ„ํ•œ ํ‘œ์ค€ ๋ฒค์น˜๋งˆํฌ๋กœ์„œ ์—ฐ๊ตฌ์ž์™€ ์—”์ง€๋‹ˆ์–ด๊ฐ€ ํ•ด๋‹น ๋ถ„์•ผ์˜ ๋ฐฉ๋ฒ•์„ ๋น„๊ตํ•˜๊ณ  ์ง„ํ–‰ ์ƒํ™ฉ์„ ์ถ”์ ํ•˜๋Š” ๋ฐ ๋„์›€์ด ๋œ๋‹ค๋Š” ๋ฐ ์žˆ์Šต๋‹ˆ๋‹ค.

Ultralytics YOLO ์„ ์‚ฌ์šฉํ•˜์—ฌ MNIST ๋ฐ์ดํ„ฐ ์„ธํŠธ์—์„œ ๋ชจ๋ธ์„ ํ›ˆ๋ จํ•˜๋ ค๋ฉด ์–ด๋–ป๊ฒŒ ํ•ด์•ผ ํ•˜๋‚˜์š”?

Ultralytics YOLO ์„ ์‚ฌ์šฉํ•˜์—ฌ MNIST ๋ฐ์ดํ„ฐ ์„ธํŠธ์—์„œ ๋ชจ๋ธ์„ ํ•™์Šต์‹œํ‚ค๋ ค๋ฉด ๋‹ค์Œ ๋‹จ๊ณ„๋ฅผ ๋”ฐ๋ฅด์„ธ์š”:

์—ด์ฐจ ์˜ˆ์‹œ

from ultralytics import YOLO

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

# Train the model
results = model.train(data="mnist", epochs=100, imgsz=32)
# Start training from a pretrained *.pt model
yolo classify train data=mnist model=yolo11n-cls.pt epochs=100 imgsz=28

์‚ฌ์šฉ ๊ฐ€๋Šฅํ•œ ๊ต์œก ์ธ์ˆ˜์— ๋Œ€ํ•œ ์ž์„ธํ•œ ๋ชฉ๋ก์€ ๊ต์œก ํŽ˜์ด์ง€๋ฅผ ์ฐธ์กฐํ•˜์„ธ์š”.

MNIST์™€ EMNIST ๋ฐ์ดํ„ฐ ์„ธํŠธ์˜ ์ฐจ์ด์ ์€ ๋ฌด์—‡์ธ๊ฐ€์š”?

MNIST ๋ฐ์ดํ„ฐ ์„ธํŠธ๋Š” ์†์œผ๋กœ ์“ด ์ˆซ์ž๋งŒ ํฌํ•จํ•˜์ง€๋งŒ, ํ™•์žฅ๋œ MNIST(EMNIST) ๋ฐ์ดํ„ฐ ์„ธํŠธ๋Š” ์ˆซ์ž์™€ ๋Œ€๋ฌธ์ž ๋ฐ ์†Œ๋ฌธ์ž๋ฅผ ๋ชจ๋‘ ํฌํ•จํ•ฉ๋‹ˆ๋‹ค. EMNIST๋Š” MNIST์˜ ํ›„์†์œผ๋กœ ๊ฐœ๋ฐœ๋˜์—ˆ์œผ๋ฉฐ ์ด๋ฏธ์ง€์— ๋™์ผํ•œ 28x28ํ”ฝ์…€ ํ˜•์‹์„ ์‚ฌ์šฉํ•˜๋ฏ€๋กœ ์›๋ž˜ MNIST ๋ฐ์ดํ„ฐ ์„ธํŠธ์šฉ์œผ๋กœ ์„ค๊ณ„๋œ ๋„๊ตฌ ๋ฐ ๋ชจ๋ธ๊ณผ ํ˜ธํ™˜๋ฉ๋‹ˆ๋‹ค. EMNIST์˜ ๋” ๋„“์€ ๋ฌธ์ž ๋ฒ”์œ„๋Š” ๋” ๋‹ค์–‘ํ•œ ๋จธ์‹  ๋Ÿฌ๋‹ ์• ํ”Œ๋ฆฌ์ผ€์ด์…˜์— ์œ ์šฉํ•ฉ๋‹ˆ๋‹ค.

Ultralytics HUB๋ฅผ ์‚ฌ์šฉํ•˜์—ฌ MNIST์™€ ๊ฐ™์€ ์‚ฌ์šฉ์ž ์ง€์ • ๋ฐ์ดํ„ฐ ์„ธํŠธ์—์„œ ๋ชจ๋ธ์„ ํ•™์Šต์‹œํ‚ฌ ์ˆ˜ ์žˆ๋‚˜์š”?

์˜ˆ, Ultralytics HUB๋ฅผ ์‚ฌ์šฉํ•˜์—ฌ MNIST์™€ ๊ฐ™์€ ์‚ฌ์šฉ์ž ์ง€์ • ๋ฐ์ดํ„ฐ ์„ธํŠธ์—์„œ ๋ชจ๋ธ์„ ํ›ˆ๋ จํ•  ์ˆ˜ ์žˆ์Šต๋‹ˆ๋‹ค. Ultralytics HUB๋Š” ๊ด‘๋ฒ”์œ„ํ•œ ์ฝ”๋”ฉ ์ง€์‹์ด ์—†์–ด๋„ ๋ฐ์ดํ„ฐ ์„ธํŠธ ์—…๋กœ๋“œ, ๋ชจ๋ธ ํ•™์Šต, ํ”„๋กœ์ ํŠธ ๊ด€๋ฆฌ๋ฅผ ์œ„ํ•œ ์‚ฌ์šฉ์ž ์นœํ™”์ ์ธ ์ธํ„ฐํŽ˜์ด์Šค๋ฅผ ์ œ๊ณตํ•ฉ๋‹ˆ๋‹ค. ์‹œ์ž‘ํ•˜๋Š” ๋ฐฉ๋ฒ•์— ๋Œ€ํ•œ ์ž์„ธํ•œ ๋‚ด์šฉ์€ Ultralytics HUB ๋น ๋ฅธ ์‹œ์ž‘ ํŽ˜์ด์ง€๋ฅผ ์ฐธ์กฐํ•˜์„ธ์š”.


๐Ÿ“… Created 11 months ago โœ๏ธ Updated 10 days ago

๋Œ“๊ธ€