์ฝ˜ํ…์ธ ๋กœ ๊ฑด๋„ˆ๋›ฐ๊ธฐ

์ผ๋ฐ˜์ ์ธ YOLO ๋ฌธ์ œ ํ•ด๊ฒฐ

YOLO ์ผ๋ฐ˜์ ์ธ ๋ฌธ์ œ ์ด๋ฏธ์ง€

์†Œ๊ฐœ

This guide serves as a comprehensive aid for troubleshooting common issues encountered while working with YOLO11 on your Ultralytics projects. Navigating through these issues can be a breeze with the right guidance, ensuring your projects remain on track without unnecessary delays.



Watch: Ultralytics YOLO11 Common Issues | Installation Errors, Model Training Issues

์ผ๋ฐ˜์ ์ธ ๋ฌธ์ œ

์„ค์น˜ ์˜ค๋ฅ˜

์„ค์น˜ ์˜ค๋ฅ˜๋Š” ํ˜ธํ™˜๋˜์ง€ ์•Š๋Š” ๋ฒ„์ „, ์ข…์†์„ฑ ๋ˆ„๋ฝ, ์ž˜๋ชป๋œ ํ™˜๊ฒฝ ์„ค์ • ๋“ฑ ๋‹ค์–‘ํ•œ ์›์ธ์œผ๋กœ ์ธํ•ด ๋ฐœ์ƒํ•  ์ˆ˜ ์žˆ์Šต๋‹ˆ๋‹ค. ๋จผ์ € ๋‹ค์Œ์„ ์ˆ˜ํ–‰ํ•˜๊ณ  ์žˆ๋Š”์ง€ ํ™•์ธํ•˜์„ธ์š”:

  • ๊ถŒ์žฅ๋˜๋Š” Python 3.8 ์ด์ƒ์„ ์‚ฌ์šฉ ์ค‘์ž…๋‹ˆ๋‹ค.

  • Ensure that you have the correct version of PyTorch (1.8 or later) installed.

  • ์ถฉ๋Œ์„ ํ”ผํ•˜๊ธฐ ์œ„ํ•ด ๊ฐ€์ƒ ํ™˜๊ฒฝ์„ ์‚ฌ์šฉํ•˜๋Š” ๊ฒƒ์„ ๊ณ ๋ คํ•˜์„ธ์š”.

  • ๊ณต์‹ ์„ค์น˜ ๊ฐ€์ด๋“œ๋ฅผ ๋‹จ๊ณ„๋ณ„๋กœ ๋”ฐ๋ผํ•˜์„ธ์š”.

๋˜ํ•œ ์‚ฌ์šฉ์ž๋“ค์ด ํ”ํžˆ ๊ฒช๋Š” ์„ค์น˜ ๋ฌธ์ œ์™€ ๊ฐ๊ฐ์˜ ํ•ด๊ฒฐ ๋ฐฉ๋ฒ•์€ ๋‹ค์Œ๊ณผ ๊ฐ™์Šต๋‹ˆ๋‹ค:

  • Import Errors or Dependency Issues - If you're getting errors during the import of YOLO11, or you're having issues related to dependencies, consider the following troubleshooting steps:

    • ์ƒˆ๋กœ ์„ค์น˜: ์ƒˆ๋กœ ์„ค์น˜ํ•˜๋ฉด ์˜ˆ๊ธฐ์น˜ ์•Š์€ ๋ฌธ์ œ๋ฅผ ํ•ด๊ฒฐํ•  ์ˆ˜ ์žˆ๋Š” ๊ฒฝ์šฐ๊ฐ€ ์žˆ์Šต๋‹ˆ๋‹ค. ํŠนํžˆ Ultralytics ์™€ ๊ฐ™์€ ๋ผ์ด๋ธŒ๋Ÿฌ๋ฆฌ์˜ ๊ฒฝ์šฐ ์—…๋ฐ์ดํŠธ๋กœ ์ธํ•ด ํŒŒ์ผ ํŠธ๋ฆฌ ๊ตฌ์กฐ๋‚˜ ๊ธฐ๋Šฅ์ด ๋ณ€๊ฒฝ๋  ์ˆ˜ ์žˆ์Šต๋‹ˆ๋‹ค.

    • ์ •๊ธฐ์ ์œผ๋กœ ์—…๋ฐ์ดํŠธํ•ฉ๋‹ˆ๋‹ค: ์ตœ์‹  ๋ฒ„์ „์˜ ๋ผ์ด๋ธŒ๋Ÿฌ๋ฆฌ๋ฅผ ์‚ฌ์šฉํ•˜๊ณ  ์žˆ๋Š”์ง€ ํ™•์ธํ•˜์„ธ์š”. ์ด์ „ ๋ฒ„์ „์€ ์ตœ์‹  ์—…๋ฐ์ดํŠธ์™€ ํ˜ธํ™˜๋˜์ง€ ์•Š์•„ ์ž ์žฌ์ ์ธ ์ถฉ๋Œ์ด๋‚˜ ๋ฌธ์ œ๊ฐ€ ๋ฐœ์ƒํ•  ์ˆ˜ ์žˆ์Šต๋‹ˆ๋‹ค.

    • ์ข…์†์„ฑ์„ ํ™•์ธํ•ฉ๋‹ˆ๋‹ค: ํ•„์š”ํ•œ ๋ชจ๋“  ์ข…์†์„ฑ์ด ์˜ฌ๋ฐ”๋ฅด๊ฒŒ ์„ค์น˜๋˜์–ด ์žˆ๊ณ  ํ˜ธํ™˜๋˜๋Š” ๋ฒ„์ „์ธ์ง€ ํ™•์ธํ•ฉ๋‹ˆ๋‹ค.

    • ๋ณ€๊ฒฝ ์‚ฌํ•ญ ๊ฒ€ํ† : ์ด์ „ ๋ฒ„์ „์„ ์ฒ˜์Œ ๋ณต์ œํ•˜๊ฑฐ๋‚˜ ์„ค์น˜ํ•œ ๊ฒฝ์šฐ, ์ค‘์š”ํ•œ ์—…๋ฐ์ดํŠธ๊ฐ€ ๋ผ์ด๋ธŒ๋Ÿฌ๋ฆฌ์˜ ๊ตฌ์กฐ๋‚˜ ๊ธฐ๋Šฅ์— ์˜ํ–ฅ์„ ๋ฏธ์น  ์ˆ˜ ์žˆ๋‹ค๋Š” ์ ์— ์œ ์˜ํ•˜์„ธ์š”. ํ•ญ์ƒ ๊ณต์‹ ๋ฌธ์„œ๋‚˜ ๋ณ€๊ฒฝ ๋กœ๊ทธ๋ฅผ ์ฐธ์กฐํ•˜์—ฌ ์ฃผ์š” ๋ณ€๊ฒฝ ์‚ฌํ•ญ์„ ํŒŒ์•…ํ•˜์„ธ์š”.

    • ์›ํ™œํ•˜๊ณ  ์˜ค๋ฅ˜ ์—†๋Š” ๊ฒฝํ—˜์„ ์œ„ํ•ด์„œ๋Š” ๋ผ์ด๋ธŒ๋Ÿฌ๋ฆฌ์™€ ์ข…์†์„ฑ์„ ์ตœ์‹  ์ƒํƒœ๋กœ ์œ ์ง€ํ•˜๋Š” ๊ฒƒ์ด ์ค‘์š”ํ•ฉ๋‹ˆ๋‹ค.

  • Running YOLO11 on GPU - If you're having trouble running YOLO11 on GPU, consider the following troubleshooting steps:

    • CUDA ํ˜ธํ™˜์„ฑ ๋ฐ ์„ค์น˜ ํ™•์ธ: GPU ์ด CUDA ํ˜ธํ™˜๋˜๋Š”์ง€, CUDA ์ด ์˜ฌ๋ฐ”๋ฅด๊ฒŒ ์„ค์น˜๋˜์—ˆ๋Š”์ง€ ํ™•์ธํ•ฉ๋‹ˆ๋‹ค. ์‚ฌ์šฉ nvidia-smi ๋ช…๋ น์„ ์‚ฌ์šฉํ•˜์—ฌ NVIDIA GPU ๋ฐ CUDA ๋ฒ„์ „์˜ ์ƒํƒœ๋ฅผ ํ™•์ธํ•  ์ˆ˜ ์žˆ์Šต๋‹ˆ๋‹ค.

    • PyTorch ๋ฐ CUDA ํ†ตํ•ฉ ํ™•์ธ: PyTorch ์„ ์‹คํ–‰ํ•˜์—ฌ CUDA ์„ ํ™œ์šฉํ•  ์ˆ˜ ์žˆ๋Š”์ง€ ํ™•์ธ import torch; print(torch.cuda.is_available()) ๋ฅผ Python ํ„ฐ๋ฏธ๋„์—์„œ ์‹คํ–‰ํ•ฉ๋‹ˆ๋‹ค. 'True'๋ฅผ ๋ฐ˜ํ™˜ํ•˜๋ฉด PyTorch ์€ CUDA ์„ ์‚ฌ์šฉํ•˜๋„๋ก ์„ค์ •๋ฉ๋‹ˆ๋‹ค.

    • ํ™˜๊ฒฝ ํ™œ์„ฑํ™”: ํ•„์š”ํ•œ ๋ชจ๋“  ํŒจํ‚ค์ง€๊ฐ€ ์„ค์น˜๋˜์–ด ์žˆ๋Š” ์˜ฌ๋ฐ”๋ฅธ ํ™˜๊ฒฝ์ธ์ง€ ํ™•์ธํ•ฉ๋‹ˆ๋‹ค.

    • ํŒจํ‚ค์ง€ ์—…๋ฐ์ดํŠธ: ์˜ค๋ž˜๋œ ํŒจํ‚ค์ง€๋Š” GPU ๊ณผ ํ˜ธํ™˜๋˜์ง€ ์•Š์„ ์ˆ˜ ์žˆ์Šต๋‹ˆ๋‹ค. ๊ณ„์† ์—…๋ฐ์ดํŠธํ•˜์„ธ์š”.

    • Program Configuration: Check if the program or code specifies GPU usage. In YOLO11, this might be in the settings or configuration.

๋ชจ๋ธ ๊ต์œก ๋ฌธ์ œ

์ด ์„น์…˜์—์„œ๋Š” ๊ต์œก ์ค‘์— ์ง๋ฉดํ•˜๋Š” ์ผ๋ฐ˜์ ์ธ ๋ฌธ์ œ์™€ ๊ทธ์— ๋Œ€ํ•œ ์„ค๋ช… ๋ฐ ํ•ด๊ฒฐ ๋ฐฉ๋ฒ•์„ ๋‹ค๋ฃน๋‹ˆ๋‹ค.

๊ตฌ์„ฑ ์„ค์ • ํ™•์ธ

์ด์Šˆ: ๊ตฌ์„ฑ ์„ค์ •์ด ์˜ฌ๋ฐ”๋ฅธ์ง€ ํ™•์‹คํ•˜์ง€ ์•Š์Šต๋‹ˆ๋‹ค. .yaml ํŒŒ์ผ์ด ๋ชจ๋ธ ํ•™์Šต ์ค‘์— ์˜ฌ๋ฐ”๋ฅด๊ฒŒ ์ ์šฉ๋˜๊ณ  ์žˆ๋Š”์ง€ ํ™•์ธํ•ฉ๋‹ˆ๋‹ค.

์†”๋ฃจ์…˜: ๊ตฌ์„ฑ ์„ค์ •์˜ .yaml ํŒŒ์ผ์„ ์‚ฌ์šฉํ•  ๋•Œ ์ ์šฉํ•ด์•ผ ํ•ฉ๋‹ˆ๋‹ค. model.train() ๊ธฐ๋Šฅ์„ ์‚ฌ์šฉํ•˜์„ธ์š”. ์ด๋Ÿฌํ•œ ์„ค์ •์ด ์˜ฌ๋ฐ”๋ฅด๊ฒŒ ์ ์šฉ๋˜์—ˆ๋Š”์ง€ ํ™•์ธํ•˜๋ ค๋ฉด ๋‹ค์Œ ๋‹จ๊ณ„๋ฅผ ๋”ฐ๋ฅด์„ธ์š”:

  • ๊ฒฝ๋กœ๊ฐ€ .yaml ๊ตฌ์„ฑ ํŒŒ์ผ์ด ์˜ฌ๋ฐ”๋ฅธ์ง€ ํ™•์ธํ•ฉ๋‹ˆ๋‹ค.
  • ๊ฒฝ๋กœ๋ฅผ ์ „๋‹ฌํ•ด์•ผ ํ•ฉ๋‹ˆ๋‹ค. .yaml ํŒŒ์ผ์„ data ์ธ์ˆ˜๋ฅผ ํ˜ธ์ถœํ•  ๋•Œ model.train()๋ฅผ ํด๋ฆญํ•ฉ๋‹ˆ๋‹ค:
model.train(data="/path/to/your/data.yaml", batch=4)

๋‹ค์ค‘ GPU๋กœ ํ›ˆ๋ จ ๊ฐ€์†ํ™”

๋ฌธ์ œ: ๋‹จ์ผ GPU ์—์„œ ํŠธ๋ ˆ์ด๋‹ ์†๋„๊ฐ€ ๋Š๋ฆฌ๊ณ  ์—ฌ๋Ÿฌ GPU๋ฅผ ์‚ฌ์šฉํ•˜์—ฌ ํ”„๋กœ์„ธ์Šค ์†๋„๋ฅผ ๋†’์ด๊ณ  ์‹ถ์Šต๋‹ˆ๋‹ค.

Solution: Increasing the batch size can accelerate training, but it's essential to consider GPU memory capacity. To speed up training with multiple GPUs, follow these steps:

  • ์—ฌ๋Ÿฌ ๊ฐœ์˜ GPU๋ฅผ ์‚ฌ์šฉํ•  ์ˆ˜ ์žˆ๋Š”์ง€ ํ™•์ธํ•ฉ๋‹ˆ๋‹ค.

  • .yaml ๊ตฌ์„ฑ ํŒŒ์ผ์„ ์ˆ˜์ •ํ•˜์—ฌ ์‚ฌ์šฉํ•  GPU ์ˆ˜(์˜ˆ: gpus)๋ฅผ ์ง€์ •ํ•ฉ๋‹ˆ๋‹ค: 4.

  • ๋ฉ”๋ชจ๋ฆฌ ์ œํ•œ์„ ์ดˆ๊ณผํ•˜์ง€ ์•Š๊ณ  ์—ฌ๋Ÿฌ GPU๋ฅผ ์ตœ๋Œ€ํ•œ ํ™œ์šฉํ•˜๋ ค๋ฉด ๋ฐฐ์น˜ ํฌ๊ธฐ๋ฅผ ์ ์ ˆํžˆ ๋Š˜๋ฆฌ์„ธ์š”.

  • ์—ฌ๋Ÿฌ GPU๋ฅผ ํ™œ์šฉํ•˜๋„๋ก ํ›ˆ๋ จ ๋ช…๋ น์„ ์ˆ˜์ •ํ•ฉ๋‹ˆ๋‹ค:

# Adjust the batch size and other settings as needed to optimize training speed
model.train(data="/path/to/your/data.yaml", batch=32, multi_scale=True)

์—ฐ์† ๋ชจ๋‹ˆํ„ฐ๋ง ๋งค๊ฐœ๋ณ€์ˆ˜

๋ฌธ์ œ: ์†์‹ค๊ณผ๋Š” ๋ณ„๋„๋กœ ํ›ˆ๋ จ ์ค‘์— ์ง€์†์ ์œผ๋กœ ๋ชจ๋‹ˆํ„ฐ๋งํ•ด์•ผ ํ•˜๋Š” ๋งค๊ฐœ๋ณ€์ˆ˜๋ฅผ ์•Œ๊ณ  ์‹ถ์Šต๋‹ˆ๋‹ค.

์†”๋ฃจ์…˜: ์†์‹ค์€ ๋ชจ๋‹ˆํ„ฐ๋งํ•ด์•ผ ํ•  ์ค‘์š”ํ•œ ์ง€ํ‘œ์ด์ง€๋งŒ, ๋ชจ๋ธ ์„ฑ๋Šฅ ์ตœ์ ํ™”๋ฅผ ์œ„ํ•ด ๋‹ค๋ฅธ ์ง€ํ‘œ๋„ ์ถ”์ ํ•˜๋Š” ๊ฒƒ์ด ํ•„์ˆ˜์ ์ž…๋‹ˆ๋‹ค. ํ›ˆ๋ จ ์ค‘์— ๋ชจ๋‹ˆํ„ฐ๋งํ•ด์•ผ ํ•  ๋ช‡ ๊ฐ€์ง€ ์ฃผ์š” ์ง€ํ‘œ๋Š” ๋‹ค์Œ๊ณผ ๊ฐ™์Šต๋‹ˆ๋‹ค:

์ด๋Ÿฌํ•œ ๋ฉ”ํŠธ๋ฆญ์€ ํŠธ๋ ˆ์ด๋‹ ๋กœ๊ทธ์—์„œ ๋˜๋Š” ์‹œ๊ฐํ™”๋ฅผ ์œ„ํ•ด TensorBoard ๋˜๋Š” wandb์™€ ๊ฐ™์€ ๋„๊ตฌ๋ฅผ ์‚ฌ์šฉํ•˜์—ฌ ์•ก์„ธ์Šคํ•  ์ˆ˜ ์žˆ์Šต๋‹ˆ๋‹ค. ์ด๋Ÿฌํ•œ ๋ฉ”ํŠธ๋ฆญ์„ ๊ธฐ๋ฐ˜์œผ๋กœ ์กฐ๊ธฐ ์ค‘์ง€๋ฅผ ๊ตฌํ˜„ํ•˜๋ฉด ๋” ๋‚˜์€ ๊ฒฐ๊ณผ๋ฅผ ์–ป์„ ์ˆ˜ ์žˆ์Šต๋‹ˆ๋‹ค.

๊ต์œก ์ง„ํ–‰ ์ƒํ™ฉ์„ ์ถ”์ ํ•˜๋Š” ๋„๊ตฌ

์ด์Šˆ: ๊ต์œก ์ง„ํ–‰ ์ƒํ™ฉ์„ ์ถ”์ ํ•  ์ˆ˜ ์žˆ๋Š” ๋„๊ตฌ์— ๋Œ€ํ•œ ์ถ”์ฒœ์„ ์ฐพ๊ณ  ์žˆ์Šต๋‹ˆ๋‹ค.

์†”๋ฃจ์…˜: ๊ต์œก ์ง„ํ–‰ ์ƒํ™ฉ์„ ์ถ”์ ํ•˜๊ณ  ์‹œ๊ฐํ™”ํ•˜๋ ค๋ฉด ๋‹ค์Œ ๋„๊ตฌ๋ฅผ ์‚ฌ์šฉํ•˜๋Š” ๊ฒƒ์„ ๊ณ ๋ คํ•ด ๋ณผ ์ˆ˜ ์žˆ์Šต๋‹ˆ๋‹ค:

  • TensorBoard: TensorBoard is a popular choice for visualizing training metrics, including loss, accuracy, and more. You can integrate it with your YOLO11 training process.
  • CometComet ์—์„œ๋Š” ์‹คํ—˜ ์ถ”์  ๋ฐ ๋น„๊ต๋ฅผ ์œ„ํ•œ ๊ด‘๋ฒ”์œ„ํ•œ ํˆดํ‚ท์„ ์ œ๊ณตํ•ฉ๋‹ˆ๋‹ค. ์ด๋ฅผ ํ†ตํ•ด ๋ฉ”ํŠธ๋ฆญ, ํ•˜์ดํผํŒŒ๋ผ๋ฏธํ„ฐ, ๋ชจ๋ธ ๊ฐ€์ค‘์น˜๊นŒ์ง€ ์ถ”์ ํ•  ์ˆ˜ ์žˆ์Šต๋‹ˆ๋‹ค. YOLO ๋ชจ๋ธ๊ณผ์˜ ํ†ตํ•ฉ๋„ ๊ฐ„๋‹จํ•˜์—ฌ ์‹คํ—˜ ์ฃผ๊ธฐ์— ๋Œ€ํ•œ ์ „์ฒด ๊ฐœ์š”๋ฅผ ์ œ๊ณตํ•ฉ๋‹ˆ๋‹ค.
  • Ultralytics ํ—ˆ๋ธŒ: Ultralytics ํ—ˆ๋ธŒ๋Š” YOLO ๋ชจ๋ธ ์ถ”์ ์— ํŠนํ™”๋œ ํ™˜๊ฒฝ์„ ์ œ๊ณตํ•˜์—ฌ ๋ฉ”ํŠธ๋ฆญ, ๋ฐ์ดํ„ฐ ์„ธํŠธ๋ฅผ ๊ด€๋ฆฌํ•˜๊ณ  ํŒ€๊ณผ ํ˜‘์—…ํ•  ์ˆ˜ ์žˆ๋Š” ์›์Šคํ†ฑ ํ”Œ๋žซํผ์„ ์ œ๊ณตํ•ฉ๋‹ˆ๋‹ค. YOLO ์— ์ดˆ์ ์„ ๋งž์ถ˜ ๋งž์ถคํ˜• ์ถ”์  ์˜ต์…˜์„ ์ œ๊ณตํ•ฉ๋‹ˆ๋‹ค.

์ด๋Ÿฌํ•œ ๋„๊ตฌ๋Š” ๊ฐ๊ฐ ๊ณ ์œ ํ•œ ์žฅ์ ์„ ์ œ๊ณตํ•˜๋ฏ€๋กœ ์„ ํƒํ•  ๋•Œ ํ”„๋กœ์ ํŠธ์˜ ํŠน์ • ์š”๊ตฌ ์‚ฌํ•ญ์„ ๊ณ ๋ คํ•  ์ˆ˜ ์žˆ์Šต๋‹ˆ๋‹ค.

๊ต์œก์ด ์ง„ํ–‰ ์ค‘์ธ์ง€ ํ™•์ธํ•˜๋Š” ๋ฐฉ๋ฒ• GPU

๋ฌธ์ œ์ž…๋‹ˆ๋‹ค: ํŠธ๋ ˆ์ด๋‹ ๋กœ๊ทธ์˜ '์žฅ์น˜' ๊ฐ’์ด 'null'์ด๋ฉฐ, GPU ์—์„œ ํŠธ๋ ˆ์ด๋‹์ด ์ง„ํ–‰๋˜๊ณ  ์žˆ๋Š”์ง€ ํ™•์‹คํ•˜์ง€ ์•Š์Šต๋‹ˆ๋‹ค.

ํ•ด๊ฒฐ ๋ฐฉ๋ฒ•: 'device' ๊ฐ’์ด 'null'์ด๋ผ๋Š” ๊ฒƒ์€ ์ผ๋ฐ˜์ ์œผ๋กœ ํŠธ๋ ˆ์ด๋‹ ํ”„๋กœ์„ธ์Šค๊ฐ€ ๊ธฐ๋ณธ ๋™์ž‘์ธ ์‚ฌ์šฉ ๊ฐ€๋Šฅํ•œ GPU ์„ ์ž๋™์œผ๋กœ ์‚ฌ์šฉํ•˜๋„๋ก ์„ค์ •๋˜์–ด ์žˆ์Œ์„ ์˜๋ฏธํ•ฉ๋‹ˆ๋‹ค. ํŠน์ • GPU ์—์„œ ๊ต์œก์ด ์ด๋ฃจ์–ด์ง€๋„๋ก ํ•˜๋ ค๋ฉด .yaml ๊ตฌ์„ฑ ํŒŒ์ผ์—์„œ 'device' ๊ฐ’์„ GPU ์ธ๋ฑ์Šค(์˜ˆ: ์ฒซ ๋ฒˆ์งธ GPU ์˜ ๊ฒฝ์šฐ '0')๋กœ ์ˆ˜๋™์œผ๋กœ ์„ค์ •ํ•  ์ˆ˜ ์žˆ์Šต๋‹ˆ๋‹ค:

device: 0

์ด๋ ‡๊ฒŒ ํ•˜๋ฉด ์ง€์ •๋œ GPU ์— ๊ต์œก ํ”„๋กœ์„ธ์Šค๊ฐ€ ๋ช…์‹œ์ ์œผ๋กœ ํ• ๋‹น๋ฉ๋‹ˆ๋‹ค. CPU ์—์„œ ๊ต์œกํ•˜๋ ค๋ฉด 'device'๋ฅผ 'cpu'๋กœ ์„ค์ •ํ•ฉ๋‹ˆ๋‹ค.

'์‹คํ–‰' ํด๋”์—์„œ ๋กœ๊ทธ์™€ ๋ฉ”ํŠธ๋ฆญ์„ ์ฃผ์‹œํ•˜์—ฌ ๊ต์œก ์ง„ํ–‰ ์ƒํ™ฉ์„ ํšจ๊ณผ์ ์œผ๋กœ ๋ชจ๋‹ˆํ„ฐ๋งํ•˜์„ธ์š”.

ํšจ๊ณผ์ ์ธ ๋ชจ๋ธ ํŠธ๋ ˆ์ด๋‹์„ ์œ„ํ•œ ์ฃผ์š” ๊ณ ๋ ค ์‚ฌํ•ญ

๋‹ค์Œ์€ ๋ชจ๋ธ ํŠธ๋ ˆ์ด๋‹๊ณผ ๊ด€๋ จ๋œ ๋ฌธ์ œ์— ์ง๋ฉดํ–ˆ์„ ๋•Œ ์—ผ๋‘์— ๋‘์–ด์•ผ ํ•  ๋ช‡ ๊ฐ€์ง€ ์‚ฌํ•ญ์ž…๋‹ˆ๋‹ค.

๋ฐ์ดํ„ฐ ์„ธํŠธ ํ˜•์‹ ๋ฐ ๋ ˆ์ด๋ธ”

  • Importance: The foundation of any machine learning model lies in the quality and format of the data it is trained on.

  • ๊ถŒ์žฅ ์‚ฌํ•ญ: ์‚ฌ์šฉ์ž ์ง€์ • ๋ฐ์ดํ„ฐ ์ง‘ํ•ฉ๊ณผ ๊ด€๋ จ ๋ ˆ์ด๋ธ”์ด ์˜ˆ์ƒ ํ˜•์‹์„ ์ค€์ˆ˜ํ•˜๋Š”์ง€ ํ™•์ธํ•˜์„ธ์š”. ์ฃผ์„์ด ์ •ํ™•ํ•˜๊ณ  ๊ณ ํ’ˆ์งˆ์ธ์ง€ ํ™•์ธํ•˜๋Š” ๊ฒƒ์ด ์ค‘์š”ํ•ฉ๋‹ˆ๋‹ค. ์ฃผ์„์ด ๋ถ€์ •ํ™•ํ•˜๊ฑฐ๋‚˜ ์ˆ˜์ค€ ์ดํ•˜์˜ ์ฃผ์„์€ ๋ชจ๋ธ์˜ ํ•™์Šต ํ”„๋กœ์„ธ์Šค๋ฅผ ๋ฐฉํ•ดํ•˜์—ฌ ์˜ˆ์ธกํ•  ์ˆ˜ ์—†๋Š” ๊ฒฐ๊ณผ๋ฅผ ์ดˆ๋ž˜ํ•  ์ˆ˜ ์žˆ์Šต๋‹ˆ๋‹ค.

๋ชจ๋ธ ์ปจ๋ฒ„์ „์Šค

  • Importance: Achieving model convergence ensures that the model has sufficiently learned from the training data.

  • Recommendation: When training a model 'from scratch', it's vital to ensure that the model reaches a satisfactory level of convergence. This might necessitate a longer training duration, with more epochs, compared to when you're fine-tuning an existing model.

Learning Rate and Batch Size

  • ์ค‘์š”์„ฑ: ์ด๋Ÿฌํ•œ ํ•˜์ดํผํŒŒ๋ผ๋ฏธํ„ฐ๋Š” ํ•™์Šต ์ค‘์— ๋ชจ๋ธ์ด ๊ฐ€์ค‘์น˜๋ฅผ ์—…๋ฐ์ดํŠธํ•˜๋Š” ๋ฐฉ๋ฒ•์„ ๊ฒฐ์ •ํ•˜๋Š” ๋ฐ ์ค‘์ถ”์ ์ธ ์—ญํ• ์„ ํ•ฉ๋‹ˆ๋‹ค.

  • ๊ถŒ์žฅ ์‚ฌํ•ญ: ์„ ํƒํ•œ ํ•™์Šต ์†๋„์™€ ๋ฐฐ์น˜ ํฌ๊ธฐ๊ฐ€ ํŠน์ • ๋ฐ์ดํ„ฐ ์„ธํŠธ์— ์ตœ์ ์ธ์ง€ ์ •๊ธฐ์ ์œผ๋กœ ํ‰๊ฐ€ํ•˜์„ธ์š”. ๋ฐ์ดํ„ฐ ์„ธํŠธ์˜ ํŠน์„ฑ๊ณผ ์ผ์น˜ํ•˜์ง€ ์•Š๋Š” ๋งค๊ฐœ๋ณ€์ˆ˜๋Š” ๋ชจ๋ธ์˜ ์„ฑ๋Šฅ์„ ์ €ํ•ดํ•  ์ˆ˜ ์žˆ์Šต๋‹ˆ๋‹ค.

ํด๋ž˜์Šค ๋ฐฐํฌ

  • ์ค‘์š”์„ฑ: ๋ฐ์ดํ„ฐ ์ง‘ํ•ฉ์˜ ํด๋ž˜์Šค ๋ถ„ํฌ๋Š” ๋ชจ๋ธ์˜ ์˜ˆ์ธก ๊ฒฝํ–ฅ์— ์˜ํ–ฅ์„ ๋ฏธ์น  ์ˆ˜ ์žˆ์Šต๋‹ˆ๋‹ค.

  • ๊ถŒ์žฅ ์‚ฌํ•ญ: ๋ฐ์ดํ„ฐ ์ง‘ํ•ฉ ๋‚ด์˜ ํด๋ž˜์Šค ๋ถ„ํฌ๋ฅผ ์ •๊ธฐ์ ์œผ๋กœ ํ‰๊ฐ€ํ•˜์„ธ์š”. ํด๋ž˜์Šค ๋ถˆ๊ท ํ˜•์ด ์žˆ๋Š” ๊ฒฝ์šฐ, ๋ชจ๋ธ์ด ๋” ๋งŽ์ด ๋ถ„ํฌํ•˜๋Š” ํด๋ž˜์Šค์— ํŽธํ–ฅ๋  ์œ„ํ—˜์ด ์žˆ์Šต๋‹ˆ๋‹ค. ์ด๋Ÿฌํ•œ ํŽธํ–ฅ์€ ํ˜ผ๋™ ํ–‰๋ ฌ์—์„œ ๋ถ„๋ช…ํ•˜๊ฒŒ ๋“œ๋Ÿฌ๋‚  ์ˆ˜ ์žˆ์œผ๋ฉฐ, ์—ฌ๊ธฐ์„œ ๋ชจ๋ธ์€ ์ฃผ๋กœ ๋‹ค์ˆ˜ ํด๋ž˜์Šค๋ฅผ ์˜ˆ์ธกํ•  ์ˆ˜ ์žˆ์Šต๋‹ˆ๋‹ค.

์‚ฌ์ „ ํ•™์Šต๋œ ๊ฐ€์ค‘์น˜๋กœ ๊ต์ฐจ ํ™•์ธ

  • ์ค‘์š”์„ฑ: ์‚ฌ์ „ ํ•™์Šต๋œ ๊ฐ€์ค‘์น˜๋ฅผ ํ™œ์šฉํ•˜๋ฉด ํŠนํžˆ ๋ฐ์ดํ„ฐ๊ฐ€ ์ œํ•œ์ ์ธ ๊ฒฝ์šฐ ๋ชจ๋ธ ํ•™์Šต์„ ์œ„ํ•œ ํ™•์‹คํ•œ ์ถœ๋ฐœ์ ์„ ์ œ๊ณตํ•  ์ˆ˜ ์žˆ์Šต๋‹ˆ๋‹ค.

  • ๊ถŒ์žฅ ์‚ฌํ•ญ: ์ง„๋‹จ ๋‹จ๊ณ„๋กœ ๋™์ผํ•œ ๋ฐ์ดํ„ฐ๋ฅผ ์‚ฌ์šฉํ•˜์—ฌ ๋ชจ๋ธ์„ ํ›ˆ๋ จํ•˜๋˜ ๋ฏธ๋ฆฌ ํ›ˆ๋ จ๋œ ๊ฐ€์ค‘์น˜๋กœ ์ดˆ๊ธฐํ™”ํ•˜๋Š” ๊ฒƒ์„ ๊ณ ๋ คํ•ด ๋ณด์„ธ์š”. ์ด ์ ‘๊ทผ ๋ฐฉ์‹์„ ํ†ตํ•ด ํ˜ผ๋™ ํ–‰๋ ฌ์ด ์ž˜ ํ˜•์„ฑ๋˜๋ฉด '์ฒ˜์Œ๋ถ€ํ„ฐ' ๋ชจ๋ธ์— ์ถ”๊ฐ€ ํ•™์Šต์ด๋‚˜ ์กฐ์ •์ด ํ•„์š”ํ•  ์ˆ˜ ์žˆ์Œ์„ ์‹œ์‚ฌํ•  ์ˆ˜ ์žˆ์Šต๋‹ˆ๋‹ค.

์ด ์„น์…˜์—์„œ๋Š” ๋ชจ๋ธ ์˜ˆ์ธก ์ค‘์— ์ง๋ฉดํ•˜๋Š” ์ผ๋ฐ˜์ ์ธ ๋ฌธ์ œ๋ฅผ ๋‹ค๋ฃน๋‹ˆ๋‹ค.

Getting Bounding Box Predictions With Your YOLO11 Custom Model

Issue: When running predictions with a custom YOLO11 model, there are challenges with the format and visualization of the bounding box coordinates.

์†”๋ฃจ์…˜:

  • Coordinate Format: YOLO11 provides bounding box coordinates in absolute pixel values. To convert these to relative coordinates (ranging from 0 to 1), you need to divide by the image dimensions. For example, let's say your image size is 640x640. Then you would do the following:
# Convert absolute coordinates to relative coordinates
x1 = x1 / 640  # Divide x-coordinates by image width
x2 = x2 / 640
y1 = y1 / 640  # Divide y-coordinates by image height
y2 = y2 / 640
  • ํŒŒ์ผ ์ด๋ฆ„: ์˜ˆ์ธก ๋Œ€์ƒ ์ด๋ฏธ์ง€์˜ ํŒŒ์ผ ์ด๋ฆ„์„ ์–ป์œผ๋ ค๋ฉด ์˜ˆ์ธก ๋ฃจํ”„ ๋‚ด์˜ ๊ฒฐ๊ณผ ๊ฐœ์ฒด์—์„œ ์ง์ ‘ ์ด๋ฏธ์ง€ ํŒŒ์ผ ๊ฒฝ๋กœ์— ์•ก์„ธ์Šคํ•ฉ๋‹ˆ๋‹ค.

Filtering Objects in YOLO11 Predictions

Issue: Facing issues with how to filter and display only specific objects in the prediction results when running YOLO11 using the Ultralytics library.

ํ•ด๊ฒฐ ๋ฐฉ๋ฒ•: ํŠน์ • ํด๋ž˜์Šค๋ฅผ ๊ฐ์ง€ํ•˜๋ ค๋ฉด classes ์ธ์ˆ˜๋ฅผ ์‚ฌ์šฉํ•˜์—ฌ ์ถœ๋ ฅ์— ํฌํ•จํ•  ํด๋ž˜์Šค๋ฅผ ์ง€์ •ํ•ฉ๋‹ˆ๋‹ค. ์˜ˆ๋ฅผ ๋“ค์–ด, ์ž๋™์ฐจ๋งŒ ๊ฐ์ง€ํ•˜๋ ค๋ฉด('cars'์˜ ํด๋ž˜์Šค ์ธ๋ฑ์Šค๊ฐ€ 2๋ผ๊ณ  ๊ฐ€์ •):

yolo task=detect mode=segment model=yolo11n-seg.pt source='path/to/car.mp4' show=True classes=2

Understanding Precision Metrics in YOLO11

Issue: Confusion regarding the difference between box precision, mask precision, and confusion matrix precision in YOLO11.

Solution: Box precision measures the accuracy of predicted bounding boxes compared to the actual ground truth boxes using IoU (Intersection over Union) as the metric. Mask precision assesses the agreement between predicted segmentation masks and ground truth masks in pixel-wise object classification. Confusion matrix precision, on the other hand, focuses on overall classification accuracy across all classes and does not consider the geometric accuracy of predictions. It's important to note that a bounding box can be geometrically accurate (true positive) even if the class prediction is wrong, leading to differences between box precision and confusion matrix precision. These metrics evaluate distinct aspects of a model's performance, reflecting the need for different evaluation metrics in various tasks.

Extracting Object Dimensions in YOLO11

Issue: Difficulty in retrieving the length and height of detected objects in YOLO11, especially when multiple objects are detected in an image.

Solution: To retrieve the bounding box dimensions, first use the Ultralytics YOLO11 model to predict objects in an image. Then, extract the width and height information of bounding boxes from the prediction results.

from ultralytics import YOLO

# Load a pre-trained YOLO11 model
model = YOLO("yolo11n.pt")

# Specify the source image
source = "https://ultralytics.com/images/bus.jpg"

# Make predictions
results = model.predict(source, save=True, imgsz=320, conf=0.5)

# Extract bounding box dimensions
boxes = results[0].boxes.xywh.cpu()
for box in boxes:
    x, y, w, h = box
    print(f"Width of Box: {w}, Height of Box: {h}")

๋ฐฐํฌ ๊ณผ์ œ

GPU ๋ฐฐํฌ ๋ฌธ์ œ

์ด์Šˆ: ๋‹ค์ค‘GPU ํ™˜๊ฒฝ์—์„œ ๋ชจ๋ธ์„ ๋ฐฐํฌํ•˜๋ฉด ์˜ˆ๊ธฐ์น˜ ์•Š์€ ๋ฉ”๋ชจ๋ฆฌ ์‚ฌ์šฉ, GPU ๊ฐ„ ์ผ๊ด€๋˜์ง€ ์•Š์€ ๊ฒฐ๊ณผ ๋“ฑ๊ณผ ๊ฐ™์€ ์˜ˆ๊ธฐ์น˜ ์•Š์€ ๋™์ž‘์ด ๋ฐœ์ƒํ•  ์ˆ˜ ์žˆ์Šต๋‹ˆ๋‹ค.

ํ•ด๊ฒฐ ๋ฐฉ๋ฒ•: ๊ธฐ๋ณธ GPU ์ดˆ๊ธฐํ™”๋ฅผ ํ™•์ธํ•ฉ๋‹ˆ๋‹ค. PyTorch ์™€ ๊ฐ™์€ ์ผ๋ถ€ ํ”„๋ ˆ์ž„์›Œํฌ๋Š” ์ง€์ •๋œ GPU๋กœ ์ „ํ™˜ํ•˜๊ธฐ ์ „์— ๊ธฐ๋ณธ GPU ์—์„œ CUDA ์ž‘์—…์„ ์ดˆ๊ธฐํ™”ํ•  ์ˆ˜ ์žˆ์Šต๋‹ˆ๋‹ค. ์˜ˆ๊ธฐ์น˜ ์•Š์€ ๊ธฐ๋ณธ ์ดˆ๊ธฐํ™”๋ฅผ ์šฐํšŒํ•˜๋ ค๋ฉด ๋ฐฐํฌ ๋ฐ ์˜ˆ์ธก ์ค‘์— GPU ์„ ์ง์ ‘ ์ง€์ •ํ•˜์„ธ์š”. ๊ทธ๋Ÿฐ ๋‹ค์Œ GPU ์‚ฌ์šฉ๋ฅ  ๋ฐ ๋ฉ”๋ชจ๋ฆฌ ์‚ฌ์šฉ๋Ÿ‰์„ ๋ชจ๋‹ˆํ„ฐ๋งํ•˜๋Š” ๋„๊ตฌ๋ฅผ ์‚ฌ์šฉํ•˜์—ฌ ์‹ค์‹œ๊ฐ„์œผ๋กœ ์ด์ƒ ์ง•ํ›„๋ฅผ ํŒŒ์•…ํ•˜์„ธ์š”. ๋˜ํ•œ ์ตœ์‹  ๋ฒ„์ „์˜ ํ”„๋ ˆ์ž„์›Œํฌ ๋˜๋Š” ๋ผ์ด๋ธŒ๋Ÿฌ๋ฆฌ๋ฅผ ์‚ฌ์šฉํ•˜๊ณ  ์žˆ๋Š”์ง€ ํ™•์ธํ•˜์„ธ์š”.

๋ชจ๋ธ ๋ณ€ํ™˜/๋‚ด๋ณด๋‚ด๊ธฐ ๋ฌธ์ œ

๋ฌธ์ œ: ๋จธ์‹  ๋Ÿฌ๋‹ ๋ชจ๋ธ์„ ๋‹ค๋ฅธ ํ˜•์‹์ด๋‚˜ ํ”Œ๋žซํผ์œผ๋กœ ๋ณ€ํ™˜ํ•˜๊ฑฐ๋‚˜ ๋‚ด๋ณด๋‚ด๋Š” ๊ณผ์ •์—์„œ ์‚ฌ์šฉ์ž์—๊ฒŒ ์˜ค๋ฅ˜๋‚˜ ์˜ˆ๊ธฐ์น˜ ์•Š์€ ๋™์ž‘์ด ๋ฐœ์ƒํ•  ์ˆ˜ ์žˆ์Šต๋‹ˆ๋‹ค.

์†”๋ฃจ์…˜:

  • ํ˜ธํ™˜์„ฑ ํ™•์ธ: ์„œ๋กœ ํ˜ธํ™˜๋˜๋Š” ๋ผ์ด๋ธŒ๋Ÿฌ๋ฆฌ ๋ฐ ํ”„๋ ˆ์ž„์›Œํฌ ๋ฒ„์ „์„ ์‚ฌ์šฉํ•˜๊ณ  ์žˆ๋Š”์ง€ ํ™•์ธํ•˜์„ธ์š”. ๋ฒ„์ „์ด ์ผ์น˜ํ•˜์ง€ ์•Š์œผ๋ฉด ๋ณ€ํ™˜ ์ค‘์— ์˜ˆ๊ธฐ์น˜ ์•Š์€ ์˜ค๋ฅ˜๊ฐ€ ๋ฐœ์ƒํ•  ์ˆ˜ ์žˆ์Šต๋‹ˆ๋‹ค.

  • ํ™˜๊ฒฝ ์žฌ์„ค์ •: Jupyter ๋˜๋Š” Colab๊ณผ ๊ฐ™์€ ๋Œ€ํ™”ํ˜• ํ™˜๊ฒฝ์„ ์‚ฌ์šฉํ•˜๋Š” ๊ฒฝ์šฐ์—๋Š” ์ค‘์š”ํ•œ ๋ณ€๊ฒฝ์ด๋‚˜ ์„ค์น˜ ํ›„ ํ™˜๊ฒฝ์„ ๋‹ค์‹œ ์‹œ์ž‘ํ•˜๋Š” ๊ฒƒ์ด ์ข‹์Šต๋‹ˆ๋‹ค. ์ƒˆ๋กœ ์‹œ์ž‘ํ•˜๋ฉด ๊ทผ๋ณธ์ ์ธ ๋ฌธ์ œ๋ฅผ ํ•ด๊ฒฐํ•  ์ˆ˜ ์žˆ๋Š” ๊ฒฝ์šฐ๊ฐ€ ์žˆ์Šต๋‹ˆ๋‹ค.

  • ๊ณต์‹ ๋ฌธ์„œ: ๋ณ€ํ™˜์— ์‚ฌ์šฉ ์ค‘์ธ ๋„๊ตฌ ๋˜๋Š” ๋ผ์ด๋ธŒ๋Ÿฌ๋ฆฌ์˜ ๊ณต์‹ ๋ฌธ์„œ๋ฅผ ํ•ญ์ƒ ์ฐธ์กฐํ•˜์„ธ์š”. ์—ฌ๊ธฐ์—๋Š” ๋ชจ๋ธ ๋‚ด๋ณด๋‚ด๊ธฐ์— ๋Œ€ํ•œ ๊ตฌ์ฒด์ ์ธ ์ง€์นจ๊ณผ ๋ชจ๋ฒ” ์‚ฌ๋ก€๊ฐ€ ํฌํ•จ๋˜์–ด ์žˆ๋Š” ๊ฒฝ์šฐ๊ฐ€ ๋งŽ์Šต๋‹ˆ๋‹ค.

  • ์ปค๋ฎค๋‹ˆํ‹ฐ ์ง€์›: ๋ผ์ด๋ธŒ๋Ÿฌ๋ฆฌ ๋˜๋Š” ํ”„๋ ˆ์ž„์›Œํฌ์˜ ๊ณต์‹ ๋ฆฌํฌ์ง€ํ† ๋ฆฌ์—์„œ ๋‹ค๋ฅธ ์‚ฌ์šฉ์ž๊ฐ€ ๋ณด๊ณ ํ•œ ์œ ์‚ฌํ•œ ๋ฌธ์ œ๋ฅผ ํ™•์ธํ•˜์„ธ์š”. ์œ ์ง€ ๊ด€๋ฆฌ์ž๋‚˜ ์ปค๋ฎค๋‹ˆํ‹ฐ์—์„œ ํ† ๋ก  ์Šค๋ ˆ๋“œ์—์„œ ํ•ด๊ฒฐ์ฑ…์ด๋‚˜ ํ•ด๊ฒฐ ๋ฐฉ๋ฒ•์„ ์ œ๊ณตํ–ˆ์„ ์ˆ˜๋„ ์žˆ์Šต๋‹ˆ๋‹ค.

  • ์ •๊ธฐ์ ์œผ๋กœ ์—…๋ฐ์ดํŠธํ•ฉ๋‹ˆ๋‹ค: ์ตœ์‹  ๋ฒ„์ „์˜ ๋„๊ตฌ ๋˜๋Š” ๋ผ์ด๋ธŒ๋Ÿฌ๋ฆฌ๋ฅผ ์‚ฌ์šฉํ•˜๊ณ  ์žˆ๋Š”์ง€ ํ™•์ธํ•˜์„ธ์š”. ๊ฐœ๋ฐœ์ž๋Š” ์•Œ๋ ค์ง„ ๋ฒ„๊ทธ๋ฅผ ์ˆ˜์ •ํ•˜๊ฑฐ๋‚˜ ๊ธฐ๋Šฅ์„ ๊ฐœ์„ ํ•˜๋Š” ์—…๋ฐ์ดํŠธ๋ฅผ ์ž์ฃผ ๋ฆด๋ฆฌ์Šคํ•ฉ๋‹ˆ๋‹ค.

  • ์ ์ง„์ ์œผ๋กœ ํ…Œ์ŠคํŠธํ•˜์„ธ์š”: ์ „์ฒด ๋ณ€ํ™˜์„ ์ˆ˜ํ–‰ํ•˜๊ธฐ ์ „์— ๋” ์ž‘์€ ๋ชจ๋ธ์ด๋‚˜ ๋ฐ์ดํ„ฐ ์„ธํŠธ๋กœ ํ”„๋กœ์„ธ์Šค๋ฅผ ํ…Œ์ŠคํŠธํ•˜์—ฌ ์ž ์žฌ์ ์ธ ๋ฌธ์ œ๋ฅผ ์กฐ๊ธฐ์— ํŒŒ์•…ํ•˜์„ธ์š”.

์ปค๋ฎค๋‹ˆํ‹ฐ ๋ฐ ์ง€์›

Engaging with a community of like-minded individuals can significantly enhance your experience and success in working with YOLO11. Below are some channels and resources you may find helpful.

๋„์›€์„ ๋ฐ›์„ ์ˆ˜ ์žˆ๋Š” ํฌ๋Ÿผ ๋ฐ ์ฑ„๋„

GitHub Issues: The YOLO11 repository on GitHub has an Issues tab where you can ask questions, report bugs, and suggest new features. The community and maintainers are active here, and it's a great place to get help with specific problems.

Ultralytics ๋””์Šค์ฝ”๋“œ ์„œ๋ฒ„: Ultralytics ๋‹ค๋ฅธ ์‚ฌ์šฉ์ž ๋ฐ ๊ฐœ๋ฐœ์ž์™€ ์†Œํ†ตํ•  ์ˆ˜ ์žˆ๋Š” ๋””์Šค์ฝ”๋“œ ์„œ๋ฒ„๊ฐ€ ์žˆ์Šต๋‹ˆ๋‹ค.

๊ณต์‹ ๋ฌธ์„œ ๋ฐ ๋ฆฌ์†Œ์Šค

Ultralytics YOLO11 Docs: The official documentation provides a comprehensive overview of YOLO11, along with guides on installation, usage, and troubleshooting.

These resources should provide a solid foundation for troubleshooting and improving your YOLO11 projects, as well as connecting with others in the YOLO11 community.

๊ฒฐ๋ก 

Troubleshooting is an integral part of any development process, and being equipped with the right knowledge can significantly reduce the time and effort spent in resolving issues. This guide aimed to address the most common challenges faced by users of the YOLO11 model within the Ultralytics ecosystem. By understanding and addressing these common issues, you can ensure smoother project progress and achieve better results with your computer vision tasks.

Ultralytics ์ปค๋ฎค๋‹ˆํ‹ฐ๋Š” ๊ท€์ค‘ํ•œ ๋ฆฌ์†Œ์Šค์ž„์„ ๊ธฐ์–ตํ•˜์„ธ์š”. ๋™๋ฃŒ ๊ฐœ๋ฐœ์ž ๋ฐ ์ „๋ฌธ๊ฐ€์™€ ์†Œํ†ตํ•˜๋ฉด ํ‘œ์ค€ ๋ฌธ์„œ์—์„œ ๋‹ค๋ฃจ์ง€ ์•Š์€ ์ถ”๊ฐ€์ ์ธ ์ธ์‚ฌ์ดํŠธ์™€ ์†”๋ฃจ์…˜์„ ์–ป์„ ์ˆ˜ ์žˆ์Šต๋‹ˆ๋‹ค. ํ•ญ์ƒ ๋ฐฐ์šฐ๊ณ , ์‹คํ—˜ํ•˜๊ณ , ๊ฒฝํ—˜์„ ๊ณต์œ ํ•˜์—ฌ ์ปค๋ฎค๋‹ˆํ‹ฐ์˜ ์ง‘๋‹จ ์ง€์‹์— ๊ธฐ์—ฌํ•˜์„ธ์š”.

ํ–‰๋ณตํ•œ ๋ฌธ์ œ ํ•ด๊ฒฐ!

์ž์ฃผ ๋ฌป๋Š” ์งˆ๋ฌธ

How do I resolve installation errors with YOLO11?

์„ค์น˜ ์˜ค๋ฅ˜๋Š” ํ˜ธํ™˜์„ฑ ๋ฌธ์ œ๋‚˜ ์ข…์†์„ฑ ๋ˆ„๋ฝ์œผ๋กœ ์ธํ•ด ๋ฐœ์ƒํ•˜๋Š” ๊ฒฝ์šฐ๊ฐ€ ๋งŽ์Šต๋‹ˆ๋‹ค. Python 3.8 ์ด์ƒ์„ ์‚ฌ์šฉํ•˜๊ณ  PyTorch 1.8 ์ด์ƒ์„ ์„ค์น˜ํ–ˆ๋Š”์ง€ ํ™•์ธํ•˜์„ธ์š”. ์ถฉ๋Œ์„ ํ”ผํ•˜๋ ค๋ฉด ๊ฐ€์ƒ ํ™˜๊ฒฝ์„ ์‚ฌ์šฉํ•˜๋Š” ๊ฒƒ์ด ์ข‹์Šต๋‹ˆ๋‹ค. ๋‹จ๊ณ„๋ณ„ ์„ค์น˜ ๊ฐ€์ด๋“œ๋Š” ๊ณต์‹ ์„ค์น˜ ๊ฐ€์ด๋“œ๋ฅผ ์ฐธ์กฐํ•˜์„ธ์š”. ๊ฐ€์ ธ์˜ค๊ธฐ ์˜ค๋ฅ˜๊ฐ€ ๋ฐœ์ƒํ•˜๋ฉด ์ƒˆ๋กœ ์„ค์น˜ํ•˜๊ฑฐ๋‚˜ ๋ผ์ด๋ธŒ๋Ÿฌ๋ฆฌ๋ฅผ ์ตœ์‹  ๋ฒ„์ „์œผ๋กœ ์—…๋ฐ์ดํŠธํ•˜์„ธ์š”.

Why is my YOLO11 model training slow on a single GPU?

๋ฐฐ์น˜ ํฌ๊ธฐ๊ฐ€ ํฌ๊ฑฐ๋‚˜ ๋ฉ”๋ชจ๋ฆฌ๊ฐ€ ๋ถ€์กฑํ•˜์—ฌ ๋‹จ์ผ GPU ์—์„œ ํ›ˆ๋ จํ•˜๋Š” ์†๋„๊ฐ€ ๋Š๋ ค์งˆ ์ˆ˜ ์žˆ์Šต๋‹ˆ๋‹ค. ํ›ˆ๋ จ ์†๋„๋ฅผ ๋†’์ด๋ ค๋ฉด ์—ฌ๋Ÿฌ ๊ฐœ์˜ GPU๋ฅผ ์‚ฌ์šฉํ•˜์„ธ์š”. ์‹œ์Šคํ…œ์— ์—ฌ๋Ÿฌ ๊ฐœ์˜ GPU๋ฅผ ์‚ฌ์šฉํ•  ์ˆ˜ ์žˆ๋Š”์ง€ ํ™•์ธํ•˜๊ณ  .yaml ๊ตฌ์„ฑ ํŒŒ์ผ์„ ์‚ฌ์šฉํ•˜์—ฌ GPU ์ˆ˜๋ฅผ ์ง€์ •ํ•  ์ˆ˜ ์žˆ์Šต๋‹ˆ๋‹ค, gpus: 4. ๋ฉ”๋ชจ๋ฆฌ ์ œํ•œ์„ ์ดˆ๊ณผํ•˜์ง€ ์•Š๊ณ  GPU๋ฅผ ์ตœ๋Œ€ํ•œ ํ™œ์šฉํ•˜๋ ค๋ฉด ๋ฐฐ์น˜ ํฌ๊ธฐ๋ฅผ ์ ์ ˆํžˆ ๋Š˜๋ฆฝ๋‹ˆ๋‹ค. ๋ช…๋ น ์˜ˆ์‹œ:

model.train(data="/path/to/your/data.yaml", batch=32, multi_scale=True)

How can I ensure my YOLO11 model is training on the GPU?

ํŠธ๋ ˆ์ด๋‹ ๋กœ๊ทธ์—์„œ 'device' ๊ฐ’์ด 'null'๋กœ ํ‘œ์‹œ๋˜๋ฉด ์ผ๋ฐ˜์ ์œผ๋กœ ํŠธ๋ ˆ์ด๋‹ ํ”„๋กœ์„ธ์Šค๊ฐ€ ์‚ฌ์šฉ ๊ฐ€๋Šฅํ•œ GPU ์„ ์ž๋™์œผ๋กœ ์‚ฌ์šฉํ•˜๋„๋ก ์„ค์ •๋˜์–ด ์žˆ์Œ์„ ์˜๋ฏธํ•ฉ๋‹ˆ๋‹ค. ํŠน์ • GPU ์„ ๋ช…์‹œ์ ์œผ๋กœ ํ• ๋‹นํ•˜๋ ค๋ฉด, ์—ฌ๋Ÿฌ๋ถ„์˜ .yaml ๊ตฌ์„ฑ ํŒŒ์ผ์„ ์ถ”๊ฐ€ํ•ฉ๋‹ˆ๋‹ค. ์˜ˆ๋ฅผ ๋“ค์–ด

device: 0

์ด๋ ‡๊ฒŒ ํ•˜๋ฉด ๊ต์œก ํ”„๋กœ์„ธ์Šค๊ฐ€ ์ฒซ ๋ฒˆ์งธ GPU ๋กœ ์„ค์ •๋ฉ๋‹ˆ๋‹ค. ๋ฅผ ์ฐธ์กฐํ•˜์‹ญ์‹œ์˜ค. nvidia-smi ๋ช…๋ น์„ ์‹คํ–‰ํ•˜์—ฌ CUDA ์„ค์ •์„ ํ™•์ธํ•ฉ๋‹ˆ๋‹ค.

How can I monitor and track my YOLO11 model training progress?

Tracking and visualizing training progress can be efficiently managed through tools like TensorBoard, Comet, and Ultralytics HUB. These tools allow you to log and visualize metrics such as loss, precision, recall, and mAP. Implementing early stopping based on these metrics can also help achieve better training outcomes.

What should I do if YOLO11 is not recognizing my dataset format?

๋ฐ์ดํ„ฐ ์„ธํŠธ์™€ ๋ ˆ์ด๋ธ”์ด ์˜ˆ์ƒ๋˜๋Š” ํ˜•์‹์„ ๋”ฐ๋ฅด๋Š”์ง€ ํ™•์ธํ•ฉ๋‹ˆ๋‹ค. ์ฃผ์„์ด ์ •ํ™•ํ•˜๊ณ  ๊ณ ํ’ˆ์งˆ์ธ์ง€ ํ™•์ธํ•ฉ๋‹ˆ๋‹ค. ๋ฌธ์ œ๊ฐ€ ๋ฐœ์ƒํ•˜๋ฉด ๋ฐ์ดํ„ฐ ์ˆ˜์ง‘ ๋ฐ ์–ด๋…ธํ…Œ์ด์…˜ ๊ฐ€์ด๋“œ์—์„œ ๋ชจ๋ฒ” ์‚ฌ๋ก€๋ฅผ ์ฐธ์กฐํ•˜์„ธ์š”. ๋ฐ์ดํ„ฐ ์ง‘ํ•ฉ์— ๋Œ€ํ•œ ์ž์„ธํ•œ ์ง€์นจ์€ ๋ฌธ์„œ์—์„œ ๋ฐ์ดํ„ฐ ์ง‘ํ•ฉ ์„น์…˜์„ ์ฐธ์กฐํ•˜์„ธ์š”.


๐Ÿ“… Created 11 months ago โœ๏ธ Updated 12 days ago

๋Œ“๊ธ€