YOLO11 ã¢ãã«ããPaddlePaddle 圢åŒã«ãšã¯ã¹ããŒãããæ¹æ³
PaddlePaddle ãæè»æ§ãããã©ãŒãã³ã¹ãåæ£ç°å¢ã§ã®äžŠååŠçæ©èœã«éç¹ã眮ãããšã§ããã®ããã»ã¹ã容æã«ããŸããããã¯ãã¹ããŒããã©ã³ããã¯ã©ãŠãããŒã¹ã®ãµãŒããŒãŸã§ãããŸããŸãªããã€ã¹ããã©ãããã©ãŒã ã§YOLO11 ã³ã³ãã¥ãŒã¿ããžã§ã³ã¢ãã«ã䜿çšã§ããããšãæå³ããŸãã
èŠããã ïŒ Ultralytics YOLO11 ã¢ãã«ãPaddlePaddle 圢åŒã«ãšã¯ã¹ããŒãããæ¹æ³ïœPaddlePaddle 圢åŒã®äž»ãªæ©èœ
PaddlePaddle ã¢ãã«åœ¢åŒãžã®ãšã¯ã¹ããŒãæ©èœã«ããã ãã¬ãŒã ã¯ãŒã¯ã§äœ¿çšããããã«ã¢ãã«ãæé©åããããšãã§ããŸãã Ultralytics YOLO11ã¢ãã«ãPaddlePaddle ãã¬ãŒã ã¯ãŒã¯ã§äœ¿çšããããã«æé©åããããšãã§ããŸããPaddlePaddle ã¯ãç£æ¥çãžã®å±éã容æã«ããããšã§ç¥ãããŠãããããŸããŸãªãã¡ã€ã³ã«ãããå®ç°å¢ã§ã³ã³ãã¥ãŒã¿ããžã§ã³ã¢ããªã±ãŒã·ã§ã³ãå±éããã®ã«é©ããéžæè¢ã§ãã
ãªãPaddlePaddle ã
ãã€ãã¥ãéçºã PaddlePaddle(PArallel Distributed Deep LEarningïŒã¯ãäžåœåã®ãªãŒãã³ãœãŒã¹ã®ãã£ãŒãã©ãŒãã³ã°ã»ãã©ãããã©ãŒã ã§ãããäž»ã«ç 究çšã«æ§ç¯ãããããã€ãã®ãã¬ãŒã ã¯ãŒã¯ãšã¯ç°ãªããPaddlePaddle ã䜿ãããããšæ¥çãè¶ ããã¹ã ãŒãºãªçµ±åãåªå ããŠããã
ã®ãããªäžè¬çãªãã¬ãŒã ã¯ãŒã¯ãšåæ§ã®ããŒã«ãšãªãœãŒã¹ãæäŸããã TensorFlowãã㊠PyTorchã®ãããªäžè¬çãªãã¬ãŒã ã¯ãŒã¯ãšåæ§ã®ããŒã«ããªãœãŒã¹ãæäŸããŠãããããããçµéšã¬ãã«ã®éçºè ãå©çšã§ããããã«ãªã£ãŠããã蟲æ¥ãå·¥å ŽãããµãŒãã¹æ¥ãŸã§ãPaddlePaddle ã®477äžäººãè¶ ãã倧èŠæš¡ãªéçºè ã³ãã¥ããã£ã¯ãAIã¢ããªã±ãŒã·ã§ã³ã®äœæãšå±éãæ¯æŽããŠããã
Ultralytics YOLO11 ã¢ãã«ãPaddlePaddle ãã©ãŒãããã«ãšã¯ã¹ããŒãããããšã§ãPaddlePaddle ã®åŒ·ã¿ã§ããããã©ãŒãã³ã¹æé©åãå©çšããããšãã§ããŸããPaddlePaddle ã¯ãå¹ççãªã¢ãã«å®è¡ãšã¡ã¢ãªäœ¿çšéã®åæžãåªå ããŸãããã®çµæãYOLO11 ã®ã¢ãã«ã¯ããã«åªããããã©ãŒãã³ã¹ãéæã§ããå¯èœæ§ããããå®çšçãªã·ããªãªã§æé«ã®çµæãåºãããšãã§ããŸãã
PaddlePaddle ã¢ãã«ã®äž»ãªç¹åŸŽ
PaddlePaddle ã¢ãã«ã¯ãå€æ§ãªå±éã·ããªãªã«å¯Ÿå¿ããæè»æ§ãããã©ãŒãã³ã¹ãã¹ã±ãŒã©ããªãã£ã«è²¢ç®ããããŸããŸãªäž»èŠæ©èœãåããŠããïŒ
-
Dynamic-to-Static Graph:PaddlePaddle ã¯ãã¢ãã«ãéçãªèšç®ã°ã©ãã«å€æã§ãããåçããéçãžã®ã³ã³ãã€ã«ããµããŒãããŠãããããã«ãããå®è¡æã®ãªãŒããŒããããåæžããæšè«æ§èœãåäžãããæé©åãå¯èœã«ãªãã
-
æŒç®åèåïŒPaddlePaddle TensorRT ã®ããã«ãæŒç®åãã¥ãŒãžã§ã³ã䜿ã£ãŠèšç®ãå¹çåãããªãŒããŒããããåæžããããã®ãã¬ãŒã ã¯ãŒã¯ã¯ãäºææ§ã®ããæŒç®ãããŒãžããããšã§ãã¡ã¢ãªè»¢éãšèšç®ã¹ããããæå°åããæšè«ãé«éåããã
-
éåå:PaddlePaddle ã¯ããã¬ãŒãã³ã°åŸã®éååãéååãèæ ®ãããã¬ãŒãã³ã°ãªã©ãéååãã¯ããã¯ããµããŒãããŠããŸãããããã®ãã¯ããã¯ã¯ããã粟床ã®äœãããŒã¿è¡šçŸã䜿çšããããšãå¯èœã«ããããã©ãŒãã³ã¹ãå¹æçã«åäžãããã¢ãã«ãµã€ãºãçž®å°ããŸãã
é åãªãã·ã§ã³PaddlePaddle
YOLO11 ã¢ãã«ãPaddlePaddle ã«ãšã¯ã¹ããŒãããããã®ã³ãŒãã«å ¥ãåã«ãPaddlePaddle ã¢ãã«ãåŸæãšããããŸããŸãªå±éã·ããªãªãèŠãŠã¿ãŸãããã
PaddlePaddle ã¯ã䜿ãããããæè»æ§ãæ§èœã®ãã©ã³ã¹ãããããç°ãªããããŸããŸãªãªãã·ã§ã³ãæäŸããŠããïŒ
-
ããã«ãµãŒãŽã£ã³ã°ïŒãã®ãã¬ãŒã ã¯ãŒã¯ã¯PaddlePaddle ãé«æ§èœãªRESTful APIãšããŠã¢ãã«ã®ãããã€ãç°¡çŽ åãããPaddle Servingã¯æ¬çªç°å¢ã«æé©ã§ãã¢ãã«ã®ããŒãžã§ã³ç®¡çããªã³ã©ã€ã³A/Bãã¹ãã倧éã®ãªã¯ãšã¹ããåŠçããã¹ã±ãŒã©ããªãã£ãªã©ã®æ©èœãæäŸããŸãã
-
ããã«æšè«APIPaddle Inference APIã¯ãã¢ãã«ã®å®è¡ãäœã¬ãã«ã§å¶åŸ¡ããããšãã§ããŸãããã®ãªãã·ã§ã³ã¯ãã«ã¹ã¿ã ã¢ããªã±ãŒã·ã§ã³å ã«ã¢ãã«ãç·å¯ã«çµ±åããå¿ èŠãããå Žåããç¹å®ã®ããŒããŠã§ã¢ã«å¯ŸããŠããã©ãŒãã³ã¹ãæé©åããå¿ èŠãããå Žåã«é©ããŠããŸãã
-
ããã«ã©ã€ãPaddle Liteã¯ããªãœãŒã¹ãéãããŠããã¢ãã€ã«æ©åšãçµã¿èŸŒã¿æ©åšãžã®å±éã®ããã«èšèšãããŠããŸããPaddle Liteã¯ãARM CPUãGPUããã®ä»ã®ç¹æ®ãªããŒããŠã§ã¢äžã§ãã¢ãã«ã®ãµã€ãºãå°ããããæšè«ãé«éåããããã«æé©åãããŠããŸãã
-
Paddle.jsïŒPaddle.jsã¯ãŠã§ããã©ãŠã¶å ã§çŽæ¥PaddlePaddle ãPaddle.jsã¯äºåã«èšç·Žãããã¢ãã«ãèªã¿èŸŒãããšããPaddle.jsãæäŸããã¢ãã«å€æããŒã«ã䜿ã£ãŠpaddle-hubããã¢ãã«ãå€æããããšãã§ããŸããWebGL/WebGPU/WebAssemblyããµããŒããããã©ãŠã¶ã§å®è¡ã§ããŸãã
PaddlePaddle ãžã®ãšã¯ã¹ããŒã ïŒYOLO11 ã¢ãã«ã®å€æ
YOLO11 ã¢ãã«ãPaddlePaddle ãã©ãŒãããã«å€æããããšã§ãå®è¡ã®æè»æ§ãåäžãããããŸããŸãªå±éã·ããªãªã«å¯ŸããŠããã©ãŒãã³ã¹ãæé©åããããšãã§ããŸãã
ã€ã³ã¹ããŒã«
å¿ èŠãªããã±ãŒãžãã€ã³ã¹ããŒã«ããã«ã¯ã以äžãå®è¡ããïŒ
ã€ã³ã¹ããŒã«ããã»ã¹ã«é¢ãã詳现ãªèª¬æãšãã¹ããã©ã¯ãã£ã¹ã«ã€ããŠã¯ãUltralytics ã€ã³ã¹ããŒã«ã¬ã€ããã芧ãã ãããYOLO11 ã«å¿ èŠãªããã±ãŒãžãã€ã³ã¹ããŒã«ããéã«ãäœããã®åé¡ãçºçããå Žåã¯ã解決çããã³ãã«ã€ããŠãããããåé¡ã¬ã€ããåç §ããŠãã ããã
䜿çšæ¹æ³
䜿ãæ¹ã®èª¬æã«å ¥ãåã«ãUltralytics YOLO11 ã®å šã¢ãã«ããšã¯ã¹ããŒãã«å¯Ÿå¿ããŠããããšã確èªããŠãããŸããã ã
䜿çšæ¹æ³
from ultralytics import YOLO
# Load the YOLO11 model
model = YOLO("yolo11n.pt")
# Export the model to PaddlePaddle format
model.export(format="paddle") # creates '/yolo11n_paddle_model'
# Load the exported PaddlePaddle model
paddle_model = YOLO("./yolo11n_paddle_model")
# Run inference
results = paddle_model("https://ultralytics.com/images/bus.jpg")
ãµããŒããããŠãããšã¯ã¹ããŒããªãã·ã§ã³ã®è©³çŽ°ã«ã€ããŠã¯ãUltralytics é 眮ãªãã·ã§ã³ã®ããã¥ã¡ã³ãããŒãžãåç §ããŠãã ããã
ãšã¯ã¹ããŒããããYOLO11 PaddlePaddle ã¢ãã«ã®å±é
Ultralytics YOLO11 ã¢ãã«ã®PaddlePaddle 圢åŒãžã®ãšã¯ã¹ããŒãã«æåãããããããããããã€ããããšãã§ããŸããPaddlePaddle ã¢ãã«ãå®è¡ããããã®æåã®ã¹ããããšããŠæšå¥šãããã®ã¯ãYOLO("./model_paddle_model") ã¡ãœããã䜿çšããããšã§ãã
ããããPaddlePaddle ã®ã¢ãã«ãä»ã®æ§ã ãªç°å¢ã§å±éããããã®è©³çŽ°ãªæé ã«ã€ããŠã¯ã以äžã®ãªãœãŒã¹ãã芧ãã ããïŒ
-
ããã«ãµãŒã:Paddle Servingã䜿ã£ãŠãPaddlePaddle ã®ã¢ãã«ãããã©ãŒãã³ããªãµãŒãã¹ãšããŠãããã€ããæ¹æ³ãåŠã³ãŸãããã
-
ããã«ã©ã€ã:Paddle Liteã䜿çšããŠãã¢ãã€ã«ããã³çµã¿èŸŒã¿ããã€ã¹äžã§ã¢ãã«ãæé©åããå±éããæ¹æ³ãæ¢ããŸãã
-
ããã«:Paddle.jsã䜿çšããŠãã¯ã©ã€ã¢ã³ããµã€ãAIçšã«Webãã©ãŠã¶ã§PaddlePaddle ã¢ãã«ãå®è¡ããæ¹æ³ãã芧ãã ããã
æŠèŠ
ãã®ã¬ã€ãã§ã¯ãUltralytics YOLO11 ã®ã¢ãã«ãPaddlePaddle ãã©ãŒãããã«ãšã¯ã¹ããŒãããããã»ã¹ã«ã€ããŠèª¬æããŸããããããã®ã¹ãããã«åŸãããšã§ãPaddlePaddle ã®åŒ·ã¿ãããŸããŸãªå±éã·ããªãªã§æŽ»çšããã¢ãã«ãããŸããŸãªããŒããŠã§ã¢ããœãããŠã§ã¢ç°å¢ã«æé©åããããšãã§ããŸãã
䜿ãæ¹ã®è©³çŽ°ã«ã€ããŠã¯ãPaddlePaddle å ¬åŒããã¥ã¡ã³ããã芧ãã ããã
Ultralytics YOLO11 ã¢ãã«ãçµ±åããæ¹æ³ããã£ãšæ€èšãããã§ããïŒç§ãã¡ã®çµ±åã¬ã€ãããŒãžã§ã¯ãæ§ã ãªãªãã·ã§ã³ãæ€èšãã貎éãªãªãœãŒã¹ãšæŽå¯ãæäŸããŸãã
ããããã質å
Ultralytics YOLO11 ã®ã¢ãã«ãPaddlePaddle ãã©ãŒãããã«ãšã¯ã¹ããŒãããã«ã¯ïŒ
Ultralytics YOLO11 ã¢ãã«ãPaddlePaddle ãã©ãŒãããã«ãšã¯ã¹ããŒãããã®ã¯ç°¡åã§ãããšã¯ã¹ããŒãããã«ã¯ export
YOLO ã¡ãœããã䜿ã£ãŠãšã¯ã¹ããŒãããŸãã以äžã¯Python ã䜿ã£ãäŸã§ãïŒ
䜿çšæ¹æ³
from ultralytics import YOLO
# Load the YOLO11 model
model = YOLO("yolo11n.pt")
# Export the model to PaddlePaddle format
model.export(format="paddle") # creates '/yolo11n_paddle_model'
# Load the exported PaddlePaddle model
paddle_model = YOLO("./yolo11n_paddle_model")
# Run inference
results = paddle_model("https://ultralytics.com/images/bus.jpg")
ãã詳现ãªã»ããã¢ãããšãã©ãã«ã·ã¥ãŒãã£ã³ã°ã«ã€ããŠã¯ãUltralytics ã€ã³ã¹ããŒã«ã¬ã€ããš å ±éã®åé¡ã¬ã€ããåç §ããŠãã ããã
ã¢ãã«ã®ãããã€ã« PaddlePaddle ã䜿çšããå©ç¹ã¯äœã§ããïŒ
PaddlePaddle ã«ã¯ãã¢ãã«å±éã«ãããŠããã€ãã®éèŠãªå©ç¹ãããïŒ
- ããã©ãŒãã³ã¹ã®æé©åïŒPaddlePaddle ã¯ãå¹ççãªã¢ãã«å®è¡ãšã¡ã¢ãªäœ¿çšéã®åæžã«åªããŠããŸãã
- ãã€ãããã¯ããã¹ã¿ãã£ãã¯ãžã®ã°ã©ãã»ã³ã³ãã€ã«ïŒåçããéçãžã®ã³ã³ãã€ã«ããµããŒãããå®è¡æã®æé©åãå¯èœã«ããã
- æŒç®åã®èåïŒäºææ§ã®ããæŒç®ãèåããããšã§ãèšç®ãªãŒããŒããããåæžããã
- éååãã¯ããã¯ïŒãã¹ããã¬ãŒãã³ã°ãšéååãèæ ®ãããã¬ãŒãã³ã°ã®äž¡æ¹ããµããŒããããã粟床ã®äœãããŒã¿è¡šçŸãå¯èœã«ããããšã§ãããã©ãŒãã³ã¹ãåäžã
Ultralytics YOLO11 ã®ã¢ãã«ãPaddlePaddle ã«ãšã¯ã¹ããŒãããããšã§ãæ§ã ãªã¢ããªã±ãŒã·ã§ã³ãããŒããŠã§ã¢ãã©ãããã©ãŒã ã§æè»æ§ãšé«ãããã©ãŒãã³ã¹ã確ä¿ããããåªããçµæãåŸãããšãã§ããŸããPaddlePaddle ã®æ©èœã«ã€ããŠã¯ããã¡ããã芧ãã ããã
YOLO11 ã¢ãã«ã®å±éã«PaddlePaddle ãéžã¶çç±ã¯ïŒ
PaddlePaddleçŸåºŠïŒãã€ãã¥ïŒã«ãã£ãŠéçºããããã®AIã¯ãç£æ¥çšããã³åæ¥çšã®AIå°å ¥ã«æé©åãããŠããããã®å€§èŠæš¡ãªéçºè ã³ãã¥ããã£ãšå ç¢ãªãã¬ãŒã ã¯ãŒã¯ã¯ãTensorFlow ãPyTorch ãšåæ§ã®åºç¯ãªããŒã«ãæäŸãããYOLO11 ã®ã¢ãã«ãPaddlePaddle ã«ãšã¯ã¹ããŒãããããšã§ãããã掻çšããããšãã§ããïŒ
- ããã©ãŒãã³ã¹ã®åäžïŒæé©ãªå®è¡é床ãšã¡ã¢ãªãããããªã³ãã®åæžã
- æè»æ§ïŒã¹ããŒããã©ã³ããã¯ã©ãŠããµãŒããŒãŸã§ãæ§ã ãªããã€ã¹ã«å¹ åºã察å¿ã
- ã¹ã±ãŒã©ããªãã£ïŒåæ£ç°å¢ã«ãããå¹ççãªäžŠååŠçæ©èœ
ãããã®æ©èœã«ãããPaddlePaddle ã¯ãYOLO11 ã¢ãã«ãæ¬çªç°å¢ã§å±éããããã®é åçãªéžæè¢ãšãªã£ãŠããã
PaddlePaddle ãä»ã®ãã¬ãŒã ã¯ãŒã¯ãšæ¯ã¹ãŠã¢ãã«ã®ããã©ãŒãã³ã¹ã¯ã©ã®ããã«åäžããã®ãïŒ
PaddlePaddle ã¢ãã«ã®æ§èœãæé©åããããã«ãããã€ãã®é«åºŠãªæè¡ãæ¡çšããŠããïŒ
- åçã°ã©ãããéçã°ã©ããžïŒå®è¡æã®æé©åã®ããã«ãã¢ãã«ãéçãªèšç®ã°ã©ãã«å€æããŸãã
- æŒç®åã®èåïŒäºææ§ã®ããæŒç®ãçµã¿åãããããšã§ãã¡ã¢ãªè»¢éãæå°éã«æããæšè«é床ãåäžãããã
- éååïŒç²ŸåºŠãç¶æããªãããã¢ãã«ãµã€ãºãçž®å°ããäœç²ŸåºŠããŒã¿ã䜿çšããŠå¹çãåäžãããŸãã
ãããã®ãã¯ããã¯ã¯å¹ççãªã¢ãã«å®è¡ãåªå ãããããPaddlePaddle ã¯ãã€ããã©ãŒãã³ã¹ãªYOLO11 ã¢ãã«ãå±éããããã®åªãããªãã·ã§ã³ãšãªã£ãŠãããæé©åã®è©³çŽ°ã«ã€ããŠã¯ãPaddlePaddle å ¬åŒããã¥ã¡ã³ããåç §ããŠãã ããã
PaddlePaddle ãYOLO11 ã¢ãã«ã«ã¯ã©ã®ãããªå±éãªãã·ã§ã³ããããŸããïŒ
PaddlePaddle ã¯æè»ãªå±éãªãã·ã§ã³ãæäŸããïŒ
- ããã«ãµãŒãïŒã¢ãã«ãRESTful APIãšããŠãããã€ããã¢ãã«ã®ããŒãžã§ãã³ã°ããªã³ã©ã€ã³A/Bãã¹ããªã©ã®æ©èœãåãããããã¯ã·ã§ã³ã«çæ³çãªãµãŒãã¹ã§ãã
- ããã«æšè«APIïŒã«ã¹ã¿ã ã¢ããªã±ãŒã·ã§ã³ã®ã¢ãã«å®è¡ãäœã¬ãã«ã§å¶åŸ¡ã
- ããã«ã©ã€ãïŒã¢ãã€ã«æ©åšãçµã¿èŸŒã¿æ©åšã®éããããªãœãŒã¹ã«ã¢ãã«ãæé©åããŸãã
- Paddle.jsïŒãŠã§ããã©ãŠã¶å ã§ã¢ãã«ãçŽæ¥ãããã€ã§ããã
ãããã®ãªãã·ã§ã³ã¯ãããã€ã¹äžã§ã®æšè«ããã¹ã±ãŒã©ãã«ãªã¯ã©ãŠããµãŒãã¹ãŸã§ãå¹ åºãå±éã·ããªãªãã«ããŒããŸããUltralytics Model Deployment Options ã®ããŒãžã§ãããå€ãã®ãããã€ã¡ã³ãæŠç¥ãã芧ãã ããã