|h+ `ddlZddlZddlmZddlmZmZddlZddl Z ddl m Z ddl m Z mZddlmZmZmZddlmZmZmZmZmZmZmZmZddlmZmZmZdd l m!Z!m"Z"dd l#m$Z$Gd d e jJZ&Gd dZ'de(fdZ)ddZ*ddZ+d de(de(de,de(de,f dZ-dZ.d!de(de(de,de(fdZ/y)"N)Path)AnyIterator)Image) dataloader distributed)GroundingDataset YOLODatasetYOLOMultiModalDataset)LOADERSLoadImagesAndVideosLoadPilAndNumpyLoadScreenshots LoadStreams LoadTensor SourceTypes autocast_list) IMG_FORMATS PIN_MEMORY VID_FORMATS)RANKcolorstr) check_filecPeZdZdZdedeffd ZdefdZdefdZ dZ d Z xZ S) InfiniteDataLoadera Dataloader that reuses workers for infinite iteration. This dataloader extends the PyTorch DataLoader to provide infinite recycling of workers, which improves efficiency for training loops that need to iterate through the dataset multiple times without recreating workers. Attributes: batch_sampler (_RepeatSampler): A sampler that repeats indefinitely. iterator (Iterator): The iterator from the parent DataLoader. Methods: __len__: Return the length of the batch sampler's sampler. __iter__: Create a sampler that repeats indefinitely. __del__: Ensure workers are properly terminated. reset: Reset the iterator, useful when modifying dataset settings during training. Examples: Create an infinite dataloader for training >>> dataset = YOLODataset(...) >>> dataloader = InfiniteDataLoader(dataset, batch_size=16, shuffle=True) >>> for batch in dataloader: # Infinite iteration >>> train_step(batch) argskwargsct||i|tj|dt |j t||_y)zHInitialize the InfiniteDataLoader with the same arguments as DataLoader. batch_samplerN)super__init__object __setattr___RepeatSamplerr__iter__iterator)selfrr __class__s U/var/www/html/test/engine/venv/lib/python3.12/site-packages/ultralytics/data/build.pyr!zInfiniteDataLoader.__init__6sA $)&)4.ASAS2TU(* returnc@t|jjS)z1Return the length of the batch sampler's sampler.)lenrsamplerr's r)__len__zInfiniteDataLoader.__len__<s4%%--..r*c#lKtt|D]}t|jyw)zICreate an iterator that yields indefinitely from the underlying iterator.N)ranger-nextr&)r'_s r)r%zInfiniteDataLoader.__iter__@s-s4y! &At}}% % &s24c t|jdsy|jjD]#}|js|j %|jj y#t $rYywxYw)zKEnsure that workers are properly terminated when the dataloader is deleted._workersN)hasattrr&r6is_alive terminate_shutdown_workers Exception)r'ws r)__del__zInfiniteDataLoader.__del__Esg 4==*5]]++ "::<KKM " MM + + -   sA0)A0,A00 A<;A<c.|j|_y)zIReset the iterator to allow modifications to the dataset during training.N) _get_iteratorr&r/s r)resetzInfiniteDataLoader.resetQs**, r*) __name__ __module__ __qualname____doc__rr!intr0rr%r=r@ __classcell__)r(s@r)rrs=0+c+S+ //&(& -r*rc(eZdZdZdefdZdefdZy)r$a* Sampler that repeats forever for infinite iteration. This sampler wraps another sampler and yields its contents indefinitely, allowing for infinite iteration over a dataset without recreating the sampler. Attributes: sampler (Dataset.sampler): The sampler to repeat. r.c||_y)zDInitialize the _RepeatSampler with a sampler to repeat indefinitely.N)r.)r'r.s r)r!z_RepeatSampler.__init__as  r*r+c#LK t|jEd{7w)z=Iterate over the sampler indefinitely, yielding its contents.N)iterr.r/s r)r%z_RepeatSampler.__iter__es#DLL) ) ) )s $"$N)rArBrCrDrr!rr%r*r)r$r$Vs*(*r*r$ worker_idctjdz}tjj |tj|y)zGSet dataloader worker seed for reproducibility across worker processes.lN)torch initial_seednprandomseed)rL worker_seeds r) seed_workerrTks1$$&.KIINN; KK r*cF|rtnt}|||j||dk(||jxs||jxsd|j xsdt ||dk(rdndt|d|j|j||dk(r|jSdS) zBBuild and return a YOLO dataset based on configuration parameters.trainNF?: ?)img_pathimgsz batch_sizeaugmenthyprectcache single_clsstridepadprefixtaskclassesdatafraction) r r r\r`rarbrErrfrgri) cfgr[batchrhmoder`rc multi_modaldatasets r)build_yolo_datasetrors'2# G ii  XX ii4>>*U6{7?C4& $ XX !% 7: r*c0t|||j||dk(||jxs||jxsd|jxsdt ||dk(rdndt |d|j|j|dk(r|jSdS) zFBuild and return a GroundingDataset based on configuration parameters.rVNFrWrXrYrZ)r[ json_filer\r]r^r_r`rarbrcrdrerfrgri) r r\r`rarbrErrfrgri)rjr[rqrkrlr`rcs r)build_groundingrrs ii  XX ii4>>*U6{7?C4& $ XX !% 7: r*rkworkersshufflerank drop_lastc t|t|}tjj }tt j t|dz|}|dk(rdntj||}tj} | jdtzt|||xr|du||tt|ddt | | S)a Create and return an InfiniteDataLoader or DataLoader for training or validation. Args: dataset (Dataset): Dataset to load data from. batch (int): Batch size for the dataloader. workers (int): Number of worker threads for loading data. shuffle (bool, optional): Whether to shuffle the dataset. rank (int, optional): Process rank in distributed training. -1 for single-GPU training. drop_last (bool, optional): Whether to drop the last incomplete batch. Returns: (InfiniteDataLoader): A dataloader that can be used for training or validation. Examples: Create a dataloader for training >>> dataset = YOLODataset(...) >>> dataloader = build_dataloader(dataset, batch=16, workers=4, shuffle=True) N)rtlUU*UU* collate_fn) rnr]rt num_workersr. pin_memoryrzworker_init_fn generatorrv)minr-rNcuda device_countos cpu_countmaxrDistributedSampler Generator manual_seedrrrgetattrrT) rnrkrsrtrurvndnwr.r~s r)build_dataloaderrs( s7| $E  "B R\\^s2qz )7 3Bbjdk&D&DWV]&^G!I -45 +GtO7L$7"  r*cd\}}}}}t|tttfrt|}|j }|j ddt tzv}|jd}|jxs|jdxs|xr| }|dk(}|r|rt|}nt|trd}nyt|ttfrt|}d}nUt|t j t"j$frd}n(t|t&j(rd}n t+d||||||fS) a Check the type of input source and return corresponding flag values. Args: source (str | int | Path | list | tuple | np.ndarray | PIL.Image | torch.Tensor): The input source to check. Returns: source (str | int | Path | list | tuple | np.ndarray | PIL.Image | torch.Tensor): The processed source. webcam (bool): Whether the source is a webcam. screenshot (bool): Whether the source is a screenshot. from_img (bool): Whether the source is an image or list of images. in_memory (bool): Whether the source is an in-memory object. tensor (bool): Whether the source is a torch.Tensor. Examples: Check a file path source >>> source, webcam, screenshot, from_img, in_memory, tensor = check_source("image.jpg") Check a webcam source >>> source, webcam, screenshot, from_img, in_memory, tensor = check_source(0) )FFFFF.ry)zhttps://zhttp://zrtsp://zrtmp://ztcp://z.streamsscreenTzZUnsupported image type. For supported types see https://docs.ultralytics.com/modes/predict) isinstancestrrErlower rpartitionrr startswith isnumericendswithrr listtuplerrrPndarrayrNTensor TypeError) sourcewebcam screenshotfrom_img in_memorytensor source_loweris_fileis_urls r) check_sourcers',7X3FJ)V&3T*+V||~ ))#.r2{[7PQ(()`a!!#^vz'B^vG]V]R]!X- g'F FG $ FT5M *v& FU[["**5 6 FELL )tuu 6:xF BBr* vid_stridebufferchannelsc$t|\}}}}}} |r |jn t|||| } | r t|} nF|r|} nA|rt ||||} n/|rt ||} n|rt ||} nt||||} t| d| | S)a_ Load an inference source for object detection and apply necessary transformations. Args: source (str | Path | torch.Tensor | PIL.Image | np.ndarray, optional): The input source for inference. batch (int, optional): Batch size for dataloaders. vid_stride (int, optional): The frame interval for video sources. buffer (bool, optional): Whether stream frames will be buffered. channels (int, optional): The number of input channels for the model. Returns: (Dataset): A dataset object for the specified input source with attached source_type attribute. Examples: Load an image source for inference >>> dataset = load_inference_source("image.jpg", batch=1) Load a video stream source >>> dataset = load_inference_source("rtsp://example.com/stream", vid_stride=2) )rrr)r)rkrr source_type) rrrrrrrr setattr) rrkrrrstreamrrrrrrns r)load_inference_sourcers*?K6>R;FFJ)V(1&$${6:W_ag7hKV$  fFU]^ !&8< !&8<%fEj[cd G]K0 Nr*)rVF F)rVFr)TryF)NrxrxF)0rrQpathlibrtypingrrnumpyrPrNPILrtorch.utils.datarrultralytics.data.datasetr r r ultralytics.data.loadersr r rrrrrrultralytics.data.utilsrrrultralytics.utilsrrultralytics.utils.checksr DataLoaderrr$rErTrorrboolrrrrKr*r)rs   4YY   HG,/6-..6-r***3,*%S%3%%TW%im%P,C^)c)3)TX)lo)r*