ããŒã¿ã»ããã®æŠèŠ
Ultralytics ã¯ãæ€åºãã€ã³ã¹ã¿ã³ã¹åå²ãããŒãºæšå®ãåé¡ãå€ãªããžã§ã¯ã远跡ãªã©ã®ã³ã³ãã¥ãŒã¿ããžã§ã³ã¿ã¹ã¯ã容æã«ããããã«ãæ§ã ãªããŒã¿ã»ããããµããŒãããŠããŸãã以äžã¯ãUltralytics ã®äž»ãªããŒã¿ã»ããã®ãªã¹ããšãåã³ã³ãã¥ãŒã¿ããžã§ã³ã¿ã¹ã¯ãšããããã®ããŒã¿ã»ããã®æŠèŠã§ãã
èŠããã ïŒ Ultralytics ããŒã¿ã»ããã®æŠèŠ
Ultralytics ãšã¯ã¹ãããŒã©ãŒ
ã³ãã¥ããã£ã»ããŒã â ïž
çŸåš ultralytics>=8.3.10
Ultralytics ãšã¯ã¹ãããŒã©ãŒã®ãµããŒãã¯å»æ¢ãããŸãããããããå¿é
ãªãïŒçŸåšã§ã¯ãåæ§ã®æ©èœãããã«åŒ·åãããæ©èœã« Ultralytics ããHUBã¯ãã¯ãŒã¯ãããŒãåçåããããã«èšèšããããçŽæçãªã³ãŒãäžèŠã®ãã©ãããã©ãŒã ã§ããUltralytics HUBã䜿ãã°ãã³ãŒããäžè¡ãæžãããšãªããããŒã¿ã®æ¢çŽ¢ãèŠèŠåã管çã楜ã«ç¶ããããšãã§ããŸãããã²ãã§ãã¯ããŠããã®åŒ·åãªæ©èœãã掻çšãã ããðã
ããŒã¿ã»ããã®åã蟌ã¿ç»åã®äœæãé¡äŒŒç»åã®æ€çŽ¢ãSQLã¯ãšãªã®å®è¡ãã»ãã³ãã£ãã¯æ€çŽ¢ã®å®è¡ãèªç¶èšèªã«ããæ€çŽ¢ãå¯èœã§ãïŒGUIã¢ããªã§å§ããããšããAPIã䜿ã£ãŠç¬èªã«æ§ç¯ããããšãã§ããŸãã詳ããã¯ãã¡ããã芧ãã ããã
- GUIãã¢ãè©Šã
- ãšã¯ã¹ãããŒã©APIã®è©³çŽ°
ç©äœæ€åº
ããŠã³ãã£ã³ã°ããã¯ã¹ãªããžã§ã¯ãæ€åºã¯ãåãªããžã§ã¯ãã®åšãã«ããŠã³ãã£ã³ã°ããã¯ã¹ãæç»ããããšã«ãããç»åå ã®ãªããžã§ã¯ããæ€åºããããŒã«ã©ã€ãºããããšãå«ãã³ã³ãã¥ãŒã¿ããžã§ã³æè¡ã§ããã
- ArgoverseïŒè±å¯ãªã¢ãããŒã·ã§ã³ãæã€éœåžç°å¢ã®3Dãã©ããã³ã°ãšã¢ãŒã·ã§ã³äºæž¬ããŒã¿ãå«ãããŒã¿ã»ããã
- COCO: Common Objects in Context (COCO)ã¯ã80ã®ãªããžã§ã¯ãã«ããŽãªãæã€å€§èŠæš¡ãªãªããžã§ã¯ãæ€åºãã»ã°ã¡ã³ããŒã·ã§ã³ããã£ãã·ã§ã³ããŒã¿ã»ããã§ããã
- LVIS: 1203ã®ãªããžã§ã¯ãã«ããŽãªãæã€å€§èŠæš¡ãªãªããžã§ã¯ãæ€åºãã»ã°ã¡ã³ããŒã·ã§ã³ããã£ãã·ã§ã³ããŒã¿ã»ããã
- COCO8: COCO trainãšCOCO valã®æåã®4æã®ç»åã®ãµãã»ããã
- COCO128: COCO trainãšCOCO valã®æåã®128æã®ç»åã®ãã¡ããã¹ãã«é©ããå°ãããµãã»ããã
- ã°ããŒãã«å°éºŠ2020Global Wheat Challenge 2020ã®ããã®å°éºŠã®é éšç»åãå«ãããŒã¿ã»ããã
- Objects365ïŒ365ã®ç©äœã«ããŽãªãš600K以äžã®æ³šéä»ãç»åãæã€ãç©äœæ€åºã®ããã®é«å質ã§å€§èŠæš¡ãªããŒã¿ã»ããã
- OpenImagesV7:Google ã«ããå æ¬çãªããŒã¿ã»ããã§ã1.7Mã®èšç·Žç»åãš42kã®æ€èšŒç»åãããã
- SKU-110KïŒ1äž1000æ以äžã®ç»åãš170äžåã®ããŠã³ãã£ã³ã°ããã¯ã¹ãå«ããå°å£²ç°å¢ã«ãããé«å¯åºŠãªç©äœæ€åºãç¹åŸŽãšããããŒã¿ã»ããã
- VisDroneïŒãããŒã³ã§æ®åœ±ããã10K以äžã®ç»åãšãããªã·ãŒã±ã³ã¹ããã®ãªããžã§ã¯ãæ€åºãšãã«ããªããžã§ã¯ããã©ããã³ã°ããŒã¿ãå«ãããŒã¿ã»ããã
- VOC: Pascal Visual Object Classes (VOC)ããŒã¿ã»ããã20ã®ãªããžã§ã¯ãã¯ã©ã¹ãš11K以äžã®ç»åãå«ãããªããžã§ã¯ãæ€åºãšã»ã°ã¡ã³ããŒã·ã§ã³ã®ããã®ããŒã¿ã»ããã
- xViewïŒ60ã®ãªããžã§ã¯ãã«ããŽãªãš100äžä»¥äžã®æ³šéä»ããªããžã§ã¯ããæã€ä¿¯ç°ç»åäžã®ãªããžã§ã¯ãæ€åºçšããŒã¿ã»ããã
- RF100ïŒå æ¬çãªã¢ãã«è©äŸ¡ã®ããã®ã7ã€ã®ç»åé åã«ãŸããã100ã®ããŒã¿ã»ãããããªãå€æ§ãªç©äœæ€åºãã³ãããŒã¯ã
- è³è «çïŒè³è «çãæ€åºããããã®ããŒã¿ã»ããã«ã¯ãè «çã®æç¡ãäœçœ®ãç¹åŸŽã«é¢ãã詳现ãå«ãMRIãŸãã¯CTã¹ãã£ã³ç»åãå«ãŸããã
- African-wildlifeïŒãããã¡ããŒããŸãŠããµã€ãã·ããŠããªã©ãã¢ããªã«ã®éçåç©ã®ç»åãéããããŒã¿ã»ããã
- 眲åïŒæ§ã ãªææžã®ç»åã«çœ²åã®æ³šéãä»ããããŒã¿ã»ããã§ãææžã®æ€èšŒãäžæ£æ€åºã®ç 究ãæ¯æŽããã
ã€ã³ã¹ã¿ã³ã¹ã®ã»ã°ã¡ã³ããŒã·ã§ã³
ã€ã³ã¹ã¿ã³ã¹ã»ã°ã¡ã³ããŒã·ã§ã³ã¯ãç»åå ã®ãªããžã§ã¯ãããã¯ã»ã«ã¬ãã«ã§èå¥ããäœçœ®ãç¹å®ããã³ã³ãã¥ãŒã¿ããžã§ã³æè¡ã§ããã
- COCO: ãªããžã§ã¯ãæ€åºãã»ã°ã¡ã³ããŒã·ã§ã³ããã£ãã·ã§ã³ä»ãã¿ã¹ã¯çšã«èšèšããã倧èŠæš¡ãªããŒã¿ã»ããã§ã20äžæ以äžã®ã©ãã«ä»ãç»åãããã
- COCO8-seg: ã€ã³ã¹ã¿ã³ã¹ã®ã»ã°ã¡ã³ããŒã·ã§ã³ã¿ã¹ã¯çšã®å°èŠæš¡ãªããŒã¿ã»ããã§ãã»ã°ã¡ã³ããŒã·ã§ã³æ³šéä»ãã®COCOç»å8æã®ãµãã»ãããå«ãã
- COCO128-seg: ã€ã³ã¹ã¿ã³ã¹ã®ã»ã°ã¡ã³ããŒã·ã§ã³ã¿ã¹ã¯çšã®å°èŠæš¡ãªããŒã¿ã»ããã§ãã»ã°ã¡ã³ããŒã·ã§ã³æ³šéä»ãã®128æã®COCOç»åã®ãµãã»ãããå«ãã
- Crack-segïŒéè·¯ãå£ã®ã²ã³å²ããæ€åºããããã«ç¹å¥ã«äœãããããŒã¿ã»ããã§ãç©äœæ€åºãšã»ã°ã¡ã³ããŒã·ã§ã³ã®äž¡æ¹ã®ã¿ã¹ã¯ã«é©çšã§ããã
- Package-seg: å庫ãç£æ¥ç°å¢ã§è·ç©ãèå¥ããããã®ããŒã¿ã»ããã§ãç©äœæ€åºãšã»ã°ã¡ã³ããŒã·ã§ã³ã®äž¡æ¹ã®ã¢ããªã±ãŒã·ã§ã³ã«é©ããŠããŸãã
- Carparts-seg: èªåè»ã®éšåãèå¥ããããã«äœãããããŒã¿ã»ããã§ãèšèšã補é ãç 究ã®ããŒãºã«å¯Ÿå¿ããŠããããªããžã§ã¯ãæ€åºãšã»ã°ã¡ã³ããŒã·ã§ã³ã®äž¡æ¹ã®ã¿ã¹ã¯ã«å¯Ÿå¿ããŠããã
ããŒãºæšå®
姿å¢æšå®ã¯ãã«ã¡ã©ãŸãã¯ã¯ãŒã«ã座æšç³»ã«å¯Ÿãããªããžã§ã¯ãã®çžå¯Ÿçãªå§¿å¢ã決å®ããããã«äœ¿çšãããæè¡ã§ãã
- COCO: ããŒãºæšå®ã¿ã¹ã¯ã®ããã«èšèšãããã人éã®ããŒãºã¢ãããŒã·ã§ã³ãå«ã倧èŠæš¡ããŒã¿ã»ããã
- COCO8-poseïŒäººéã®ããŒãºã¢ãããŒã·ã§ã³ãä»ãã8æã®COCOç»åã®ãµãã»ãããå«ããããŒãºæšå®ã¿ã¹ã¯çšã®å°èŠæš¡ããŒã¿ã»ããã
- Tiger-pose: ãã©ã«çŠç¹ãåœãŠã263æã®ç»åãããªãã³ã³ãã¯ããªããŒã¿ã»ããã§ãããŒãºæšå®ã¿ã¹ã¯ã®ããã«ãã©1é ã«ã€ã12åã®ããŒãã€ã³ããã¢ãããŒã·ã§ã³ãããŠããã
- æã®ããŒãã€ã³ãïŒäººéã®æãäžå¿ãšãã26,000ç¹ä»¥äžã®ç»åããæ§æãããç°¡æœãªããŒã¿ã»ããã§ã1ã€ã®æã«ã€ã21ã®ããŒãã€ã³ããã¢ãããŒã·ã§ã³ãããŠãããããŒãºæšå®ã¿ã¹ã¯çšã«èšèšãããŠããŸãã
- Dog-pose: ç¬ã«çŠç¹ãåœãŠãçŽ6,000æã®ç»åãããªãå æ¬çãªããŒã¿ã»ããã§ãç¬1é ã«ã€ã24ã®ããŒãã€ã³ããã¢ãããŒã·ã§ã³ãããŠãããããŒãºæšå®ã¿ã¹ã¯çšã«èª¿æŽãããŠããã
åé¡
ç»ååé¡ã¯ãç»åããã®èŠèŠçå 容ã«åºã¥ããŠ1ã€ãŸãã¯è€æ°ã®å®çŸ©æžã¿ã®ã¯ã©ã¹ãŸãã¯ã«ããŽãªã«åé¡ããããšãå«ãã³ã³ãã¥ãŒã¿ããžã§ã³ã®ã¿ã¹ã¯ã§ãã
- Caltech 101: ç»ååé¡ã¿ã¹ã¯ã®ããã®101ã®ãªããžã§ã¯ãã«ããŽãªã®ç»åãå«ãããŒã¿ã»ããã
- ã«ã«ããã¯256Caltech101ã®æ¡åŒµçã§ã256ã®ãªããžã§ã¯ãã«ããŽãªãŒãšãããé£æ床ã®é«ãç»åãçšæãããŠããã
- CIFAR-10: 32x32ã®ã«ã©ãŒç»å60Kæã10ã¯ã©ã¹ã«åé¡ããããŒã¿ã»ããã
- CIFAR-100ïŒCIFAR-10ã®æ¡åŒµçã§ã1ã¯ã©ã¹ããã100ã®ãªããžã§ã¯ãã«ããŽãªãš600ã®ç»åãæã€ã
- Fashion-MNISTïŒç»ååé¡ã¿ã¹ã¯ã®ããã®ã10ã®ãã¡ãã·ã§ã³ã«ããŽãªã®70,000ã°ã¬ãŒã¹ã±ãŒã«ç»åãããªãããŒã¿ã»ããã
- ImageNetïŒ1,400äžä»¥äžã®ç»åãš20,000以äžã®ã«ããŽãªãæã€ãç©äœæ€åºãšç»ååé¡ã®ããã®å€§èŠæš¡ãªããŒã¿ã»ããã
- ImageNet-10ïŒå®éšãšãã¹ããããè¿ éã«è¡ãããã®ãImageNetã®10ã«ããŽãªããå°ãããµãã»ããã
- ImagenetteïŒImageNetã®å°ããªãµãã»ããã§ã10åã®åºå¥ããããã¯ã©ã¹ãå«ãŸããèšç·Žãšãã¹ããè¿ éã«è¡ãããšãã§ããã
- ImagewoofïŒç»ååé¡ã¿ã¹ã¯ã®ããã®10ç¬çš®ã«ããŽãªãå«ããImageNetã®ããå°é£ãªãµãã»ããã
- MNISTïŒææžãæ°åã®ã°ã¬ãŒã¹ã±ãŒã«ç»å70,000æãããªãç»ååé¡çšããŒã¿ã»ããã
- MNIST160ïŒMNISTããŒã¿ã»ãããããåMNISTã«ããŽãªã®æåã®8ç»åãããŒã¿ã»ããã«ã¯åèš160æã®ç»åãå«ãŸããã
ãªãªãšã³ãããã»ããŠã³ãã£ã³ã°ã»ããã¯ã¹ïŒOBBïŒ
OBBïŒOriented Bounding BoxesïŒã¯ãå転ããããŠã³ãã£ã³ã°ããã¯ã¹ã䜿çšããŠç»åå ã®è§åºŠã®ãããªããžã§ã¯ããæ€åºããããã®ã³ã³ãã¥ãŒã¿ããžã§ã³ã®ææ³ã§ãããå€ãã®å Žåãèªç©ºç»åãè¡æç»åã«é©çšãããã
- DOTA-v2ïŒ170äžåã®ã€ã³ã¹ã¿ã³ã¹ãš11,268æã®ç»åãæã€äººæ°ã®OBBèªç©ºç»åããŒã¿ã»ããã
- DOTA8: DOTAv1ã¹ããªããã»ããã®æåã®8æã®ç»åïŒãã¬ãŒãã³ã°çš4æãæ€èšŒçš4æïŒã®ãµãã»ããã
ãã«ããªããžã§ã¯ãã»ãã©ããã³ã°
ãã«ããªããžã§ã¯ããã©ããã³ã°ã¯ããããªã·ãŒã±ã³ã¹å ã®è€æ°ã®ãªããžã§ã¯ããæ€åºããæéçµéãšãšãã«è¿œè·¡ããã³ã³ãã¥ãŒã¿ããžã§ã³æè¡ã§ããã
- ArgoverseïŒå€ãªããžã§ã¯ã远跡ã¿ã¹ã¯ã®ããã®è±å¯ãªã¢ãããŒã·ã§ã³ãæã€ãéœåžç°å¢ããã®3D远跡ããã³éåäºæž¬ããŒã¿ãå«ãããŒã¿ã»ããã
- VisDroneïŒãããŒã³ã§æ®åœ±ããã10K以äžã®ç»åãšãããªã·ãŒã±ã³ã¹ããã®ãªããžã§ã¯ãæ€åºãšãã«ããªããžã§ã¯ããã©ããã³ã°ããŒã¿ãå«ãããŒã¿ã»ããã
æ°ããããŒã¿ã»ãããæäŸãã
æ°ããããŒã¿ã»ãããæäŸããã«ã¯ãæ¢åã®ã€ã³ãã©ãšããŸãæŽåãããããã®ããã€ãã®ã¹ããããå¿ èŠã§ããã以äžã«å¿ èŠãªã¹ãããã瀺ãïŒ
æ°ããããŒã¿ã»ãããæçš¿ããæé
- ç»åãéããïŒããŒã¿ã»ããã«å±ããç»åãéããããããã®ç»åã¯ãå ¬å ±ã®ããŒã¿ããŒã¹ãããªãèªèº«ã®ã³ã¬ã¯ã·ã§ã³ãªã©ãããŸããŸãªæ å ±æºããéããããšãã§ããã
- ç»åã«æ³šéãä»ããïŒã¿ã¹ã¯ã«å¿ããŠããããã®ç»åã«ããŠã³ãã£ã³ã°ããã¯ã¹ãã»ã°ã¡ã³ãããŸãã¯ããŒãã€ã³ãã§æ³šéãä»ããŸãã
- 泚éã®ãšã¯ã¹ããŒã:ãããã®æ³šéãYOLO
*.txt
Ultralytics ã -
ããŒã¿ã»ããã®æŽç:ããŒã¿ã»ãããæ£ãããã©ã«ãæ§é ã«æŽçããŠãã ããããã®é
train/
ãããŠval/
ãããã¬ãã«ã»ãã£ã¬ã¯ããªããããããããã®äžã«images/
ãããŠlabels/
ãµããã£ã¬ã¯ããªã«ããã -
ãäœæããã
data.yaml
ãã¡ã€ã«:ããŒã¿ã»ããã®ã«ãŒãã»ãã£ã¬ã¯ããªã«data.yaml
ãã¡ã€ã«ã«ã¯ãããŒã¿ã»ãããã¯ã©ã¹ããã®ä»å¿ èŠãªæ å ±ãèšè¿°ãããŠããã - ç»åã®æé©åïŒãªãã·ã§ã³ïŒïŒããå¹ççãªåŠçã®ããã«ããŒã¿ã»ããã®ãµã€ãºãå°ãããããå Žåã¯ã以äžã®ã³ãŒãã䜿çšããŠç»åãæé©åããããšãã§ããŸããããã¯å¿ é ã§ã¯ãããŸããããããŒã¿ã»ããã®ãµã€ãºãå°ããããããŠã³ããŒãé床ãéãããããã«ãå§ãããŸãã
- ããŒã¿ã»ãããZIPå§çž®ããïŒããŒã¿ã»ãããã©ã«ãå šäœãzipãã¡ã€ã«ã«å§çž®ããã
- ããã¥ã¡ã³ããšPRããªãã®ããŒã¿ã»ãããšããããæ¢åã®ãã¬ãŒã ã¯ãŒã¯ã«ã©ã®ããã«é©åãããã説æããããã¥ã¡ã³ãããŒãžãäœæããããã®åŸãPull Request (PR)ãæåºãããPRã®æåºæ¹æ³ã®è©³çŽ°ã«ã€ããŠã¯ãUltralytics Contribution Guidelinesãåç §ããŠãã ããã
ããŒã¿ã»ãããæé©åããŠå§çž®ããã³ãŒãäŸ
ããŒã¿ã»ããã®æé©åãšå§çž®
from pathlib import Path
from ultralytics.data.utils import compress_one_image
from ultralytics.utils.downloads import zip_directory
# Define dataset directory
path = Path("path/to/dataset")
# Optimize images in dataset (optional)
for f in path.rglob("*.jpg"):
compress_one_image(f)
# Zip dataset into 'path/to/dataset.zip'
zip_directory(path)
ãããã®ã¹ãããã«åŸãããšã§ãUltralytics' æ¢åã®æ§é ãšããŸãçµ±åããæ°ããããŒã¿ã»ãããæäŸããããšãã§ããã
ããããã質å
Ultralytics ãã©ã®ãããªããŒã¿ã»ãããç©äœæ€åºã«å¯Ÿå¿ããŠããŸããïŒ
Ultralytics ãå«ããç©äœæ€åºã®ããã®å€çš®å€æ§ãªããŒã¿ã»ããããµããŒãããŠããïŒ
- COCO: 80ã®ãªããžã§ã¯ãã«ããŽãªãæã€å€§èŠæš¡ãªãªããžã§ã¯ãæ€åºãã»ã°ã¡ã³ããŒã·ã§ã³ããã£ãã·ã§ã³ããŒã¿ã»ããã
- LVIS: 1203ã®ãªããžã§ã¯ãã«ããŽãªãæã€åºç¯ãªããŒã¿ã»ããã§ããã现ãããªããžã§ã¯ãæ€åºãšã»ã°ã¡ã³ããŒã·ã§ã³ã®ããã«èšèšãããŠããã
- ArgoverseïŒè±å¯ãªã¢ãããŒã·ã§ã³ãæã€éœåžç°å¢ã®3Dãã©ããã³ã°ãšã¢ãŒã·ã§ã³äºæž¬ããŒã¿ãå«ãããŒã¿ã»ããã
- VisDroneïŒãããŒã³ã§æ®åœ±ãããç»åããã®ç©äœæ€åºãšè€æ°ç©äœã®è¿œè·¡ããŒã¿ãå«ãããŒã¿ã»ããã
- SKU-110KïŒ1äž1000æãè¶ ããç»åã§ãå°å£²ç°å¢ã«ãããé«å¯åºŠãªç©äœæ€åºãå®çŸã
ãããã®ããŒã¿ã»ããã¯ãæ§ã ãªç©äœæ€åºã¢ããªã±ãŒã·ã§ã³ã®ããã®ããã¹ãã¢ãã«ã®åŠç¿ã容æã«ããã
æ°ããããŒã¿ã»ãããUltralytics ã«æçš¿ããã«ã¯ïŒ
æ°ããããŒã¿ã»ãããæäŸããã«ã¯ãããã€ãã®ã¹ããããããïŒ
- ç»åãéããïŒå ¬å ±ã®ããŒã¿ããŒã¹ãå人ã®ã³ã¬ã¯ã·ã§ã³ããç»åãéããã
- ç»åã«æ³šéãä»ããïŒã¿ã¹ã¯ã«å¿ããŠãããŠã³ãã£ã³ã°ããã¯ã¹ãã»ã°ã¡ã³ãããŸãã¯ããŒãã€ã³ããé©çšããŸãã
- 泚éã®ãšã¯ã¹ããŒã:ã¢ãããŒã·ã§ã³ãYOLO
*.txt
ãšãã圢åŒããšã£ãŠããã - ããŒã¿ã»ããã®æŽç:ã§ãã©ã«ãæ§é ã䜿çšããã
train/
ãããŠval/
ãã£ã¬ã¯ããªããããããããã«images/
ãããŠlabels/
ãµããã£ã¬ã¯ããªã«ããã - ãäœæããã
data.yaml
ãã¡ã€ã«:ããŒã¿ã»ããã®èª¬æãã¯ã©ã¹ããã®ä»ã®é¢é£æ å ±ãå«ãã - ç»åã®æé©åïŒãªãã·ã§ã³ïŒïŒå¹çåã®ããã«ããŒã¿ã»ãããµã€ãºãçž®å°ããŸãã
- ããŒã¿ã»ãããZIPå§çž®ããïŒããŒã¿ã»ãããzipãã¡ã€ã«ã«å§çž®ããã
- ææžãšPRããªãã®ããŒã¿ã»ããã説æããUltralytics Contribution Guidelinesã«åŸã£ãŠPull RequestãæåºããŠãã ããã
å æ¬çãªã¬ã€ãã¯Contribute New Datasetsãã芧ãã ããã
èªåã®ããŒã¿ã»ããã«Ultralytics Explorer ã䜿ãã¹ãçç±ã¯ïŒ
Ultralytics ãšã¯ã¹ãããŒã©ãŒã¯ãããŒã¿ã»ããåæã®ããã®åŒ·åãªæ©èœãæäŸããŠããŸãïŒ
- åã蟌ã¿çæïŒç»åã®ãã¯ãã«åã蟌ã¿ãäœæããŸãã
- ã»ãã³ãã£ãã¯æ€çŽ¢ïŒåã蟌ã¿ãAIã䜿ã£ãŠé¡äŒŒç»åãæ€çŽ¢ã
- SQLã¯ãšãªïŒè©³çŽ°ãªããŒã¿åæã®ããã«é«åºŠãªSQLã¯ãšãªãå®è¡ããŸãã
- èªç¶èšèªæ€çŽ¢ïŒäœ¿ãããããè¿œæ±ããèªç¶èšèªã«ããæ€çŽ¢ã
詳ããã¯Ultralytics ãšã¯ã¹ãããŒã©ãŒã§ GUIãã¢ããè©Šããã ããã
Ultralytics YOLO ã³ã³ãã¥ãŒã¿ããžã§ã³çšã¢ãã«ã®ãŠããŒã¯ãªç¹åŸŽã¯ïŒ
Ultralytics YOLO ã¢ãã«ã«ã¯ããã€ãã®ãŠããŒã¯ãªç¹åŸŽãããïŒ
- ãªã¢ã«ã¿ã€ã æ§èœïŒé«éæšè«ãšãã¬ãŒãã³ã°ã
- æ±çšæ§ïŒæ€åºãã»ã°ã¡ã³ããŒã·ã§ã³ãåé¡ã姿å¢æšå®ã¿ã¹ã¯ã«é©ããŠããã
- äºååŠç¿æžã¿ã¢ãã«ïŒæ§ã ãªã¢ããªã±ãŒã·ã§ã³ã«å¯Ÿå¿ããé«æ§èœã®äºååŠç¿æžã¿ã¢ãã«ã«ã¢ã¯ã»ã¹ã§ããŸãã
- åºç¯ãªã³ãã¥ããã£ã»ãµããŒãïŒæŽ»çºãªã³ãã¥ããã£ãšããã©ãã«ã·ã¥ãŒãã£ã³ã°ãéçºã®ããã®å æ¬çãªããã¥ã¡ã³ãã
YOLO ã Ultralytics YOLOããŒãžãã芧ãã ããã
Ultralytics ããŒã«ã䜿ã£ãŠããŒã¿ã»ãããæé©åããzipå§çž®ããã«ã¯ã©ãããã°ããã§ããïŒ
Ultralytics ããŒã«ã䜿ã£ãŠããŒã¿ã»ãããæé©åããzipå§çž®ããã«ã¯ã次ã®ã³ãŒãäŸã«åŸã£ãŠãã ããïŒ
ããŒã¿ã»ããã®æé©åãšå§çž®
from pathlib import Path
from ultralytics.data.utils import compress_one_image
from ultralytics.utils.downloads import zip_directory
# Define dataset directory
path = Path("path/to/dataset")
# Optimize images in dataset (optional)
for f in path.rglob("*.jpg"):
compress_one_image(f)
# Zip dataset into 'path/to/dataset.zip'
zip_directory(path)
ããŒã¿ã»ããã®æé©åãšZipå§çž®ã®æ¹æ³ã«ã€ããŠã¯ããã¡ããã芧ãã ããã