/j*ddlmZddlZddlZddlmZddlmZddlZddl Z ddl m Z ddl m Z mZddlmZddlmZddlmZdd lmZmZmZdd lmZdd lmZmZdd lmZm Z Gd deZ!dS)) annotationsN)copy)Any)build_dataloaderbuild_yolo_dataset) BaseTrainer)yolo)DetectionModel) DEFAULT_CFGLOGGERRANK)override_configs) plot_images plot_labels)torch_distributed_zero_first unwrap_modelceZdZdZeddfd-fd Zd.d/dZd0d1dZd2dZdZ d3d4d Z d!Z d5d6d%Z d&Z d7d*Zd+Zfd,ZxZS)8DetectionTraineraA class extending the BaseTrainer class for training based on a detection model. This trainer specializes in object detection tasks, handling the specific requirements for training YOLO models for object detection including dataset building, data loading, preprocessing, and model configuration. Attributes: model (DetectionModel): The YOLO detection model being trained. data (dict): Dictionary containing dataset information including class names and number of classes. loss_names (tuple): Names of the loss components used in training (box_loss, cls_loss, dfl_loss). Methods: build_dataset: Build YOLO dataset for training or validation. get_dataloader: Construct and return dataloader for the specified mode. preprocess_batch: Preprocess a batch of images by scaling and converting to float. set_model_attributes: Set model attributes based on dataset information. get_model: Return a YOLO detection model. get_validator: Return a validator for model evaluation. label_loss_items: Return a loss dictionary with labeled training loss items. progress_string: Return a formatted string of training progress. plot_training_samples: Plot training samples with their annotations. plot_training_labels: Create a labeled training plot of the YOLO model. auto_batch: Calculate optimal batch size based on model memory requirements. Examples: >>> from ultralytics.models.yolo.detect import DetectionTrainer >>> args = dict(model="yolo26n.pt", data="coco8.yaml", epochs=3) >>> trainer = DetectionTrainer(overrides=args) >>> trainer.train() N overridesdict[str, Any] | None _callbacks dict | NonecNt|||dS)aInitialize a DetectionTrainer object for training YOLO object detection models. Args: cfg (dict, optional): Default configuration dictionary containing training parameters. overrides (dict, optional): Dictionary of parameter overrides for the default configuration. _callbacks (dict, optional): Dictionary of callback functions to be executed during training. N)super__init__)selfcfgrr __class__s i/home/longshao/multi-rider-rag/.venv/lib/python3.11/site-packages/ultralytics/models/yolo/detect/train.pyrzDetectionTrainer.__init__7s' i44444trainimg_pathstrmodebatch int | Nonec ttt|jjd}t |j|||j||dk|S)aBuild YOLO Dataset for training or validation. Args: img_path (str): Path to the folder containing images. mode (str): 'train' mode or 'val' mode, users are able to customize different augmentations for each mode. batch (int, optional): Size of batches, this is for 'rect' mode. Returns: (Dataset): YOLO dataset object configured for the specified mode. val)r$rectstride)maxintrmodelr+rargsdata)rr"r$r%gss r build_datasetzDetectionTrainer.build_datasetAs]\$*--488::;;R @ @!$)XudidY]afYfoqrrrrr r dataset_path batch_sizer-rankc |dvsJd|dt|5||||}dddn #1swxYwY|dk}t|ddr@|r>tj|j|jdkst jd d}t|||dkr |j j n|j j d z|||j j o|dk S) aConstruct and return dataloader for the specified mode. Args: dataset_path (str): Path to the dataset. batch_size (int): Number of images per batch. rank (int): Process rank for distributed training. mode (str): 'train' for training dataloader, 'val' for validation dataloader. Returns: (DataLoader): PyTorch dataloader object. >r)r!z#Mode must be 'train' or 'val', not .Nr!r*FrzJ'rect=True' is incompatible with DataLoader shuffle, setting shuffle=False)r%workersshuffler6 drop_last) rr2getattrnpall batch_shapesr warningrr/r:compile)rr4r5r6r$datasetr;s rget_dataloaderzDetectionTrainer.get_dataloaderOsf'''')Vt)V)V)V''' )$ / / I I((tZHHG I I I I I I I I I I I I I I I'/ 7FE * * w rvgFZ^e^rst^uFu?v?v  Ng h h hG )-DI%%di>ORS>Si';DGO     sAA  A dictreturnc|D]L\}}t|tjr-|jjjdk||<M|ddz |d<jj dkr|d}tj tjj djj z ztjj djj zzjzjzjz}|t|jddz d kr?fd |jddD}t"j||d d }||d<|S)zPreprocess a batch of images by scaling and converting to float. Args: batch (dict): Dictionary containing batch data with 'img' tensor. Returns: (dict): Preprocessed batch with normalized images. cuda) non_blockingimggg?r9Nc^g|])}tj|zjz jz*S)mathceilr+).0xrsfs r z5DetectionTrainer.preprocess_batch..sBFGDIa"ft{233dkAr bilinearF)sizer$ align_corners)items isinstancetorchTensortodevicetypefloatr/ multi_scalerandom randranger-imgszr+r,shapenn functional interpolate)rr%kvimgssznsrSs` @rpreprocess_batchz!DetectionTrainer.preprocess_batchksKKMM V VDAq!U\** V44 $+:Jf:T4UUaU|))++c1e 9 3 & &}00BZ_d0eeE%L r c|jd|j_|jd|j_|j|j_t |jdr'|j|jjdSdS)z2Set model attributes based on dataset information.ncnamesend2end)max_detN)r0r.rorpr/r= set_head_attrrrrs rset_model_attributesz%DetectionTrainer.set_model_attributesst  $ 9W- )  4:y ) ) @ J $ $TY-> $ ? ? ? ? ? @ @r Tr str | Noneweightsverboseboolct||jd|jd|o tdk}|r|||S)a=Return a YOLO detection model. Args: cfg (str, optional): Path to model configuration file. weights (str, optional): Path to model weights. verbose (bool): Whether to display model information. Returns: (DetectionModel): YOLO detection model. rochannels)rochrx)r r0r load)rrrwrxr.s r get_modelzDetectionTrainer.get_modelsSsty49Z;PZaZpfjnpfpqqq  JJw    r cd|_tj|j|jt |j|jS)z6Return a DetectionValidator for YOLO model validation.)box_losscls_lossdfl_loss)save_dirr/r) loss_namesr detectDetectionValidator test_loaderrrr/ callbacksrts r get_validatorzDetectionTrainer.get_validatorsB<{--  t}4 ??W[We.   r loss_itemslist[float] | Noneprefixcfd|jD}|)d|D}tt||S|S)a_Return a loss dict with labeled training loss items tensor. Args: loss_items (list[float], optional): List of loss values. prefix (str): Prefix for keys in the returned dictionary. Returns: (dict | list): Dictionary of labeled loss items if loss_items is provided, otherwise list of keys. cg|] }d| S)/rN)rQrRrs rrTz5DetectionTrainer.label_loss_items..s#999A6A999r NcHg|]}tt|d S))roundr_)rQrRs rrTz5DetectionTrainer.label_loss_items..s(AAA%a!,,AAAr )rrEzip)rrrkeyss ` rlabel_loss_itemsz!DetectionTrainer.label_loss_itemssV:999999  !AAjAAAJD*--.. .Kr c^dddt|jzzzddg|jddRzS)z`Return a formatted string of training progress with epoch, GPU memory, loss, instances and size. z%11sEpochGPU_mem InstancesSize)lenrrts rprogress_stringz DetectionTrainer.progress_stringsXvS%9%9!9::  ? _?   ?  ? ?   r dict[str, Any]niNonecZt||d|jd|dz |jdS)zPlot training samples with their annotations. Args: batch (dict[str, Any]): Dictionary containing batch data. ni (int): Batch index used for naming the output file. im_file train_batchz.jpg)labelspathsfnameon_plotN)rrr)rr%rs rplot_training_samplesz&DetectionTrainer.plot_training_samplessK  "-"8"8"8"88L       r c4tjd|jjjDd}tjd|jjjDd}t |||jd|j|j dS)z1Create a labeled training plot of the YOLO model.cg|] }|d S)bboxesrNrQlbs rrTz9DetectionTrainer.plot_training_labels..sXXX8 XXXr rcg|] }|d S)clsrNrs rrTz9DetectionTrainer.plot_training_labels..sSSSBbiSSSr rp)rprrN) r> concatenate train_loaderrCrrsqueezer0rr)rboxesrs rplot_training_labelsz%DetectionTrainer.plot_training_labelssXXt7H7P7WXXXZ[\\nSS$2C2K2RSSSUVWWE3;;== '0BT]dhdpqqqqqqr c:t|jddi5|_||jddd}dddn #1swxYwYt d|jDd z}~t |S) zGet optimal batch size by calculating memory occupation of model. Returns: (int): Optimal batch size. cacheF)rr!r3)r$r%Nc3@K|]}t|dVdS)rN)r)rQlabels r z.DetectionTrainer.auto_batch..s.NN#eEl++NNNNNNr r)rr/r2r0r,rr auto_batch)r train_dataset max_num_objrs rrzDetectionTrainer.auto_batchs diGU3C D D D [  ..ty/AWY.ZZM [ [ [ [ [ [ [ [ [ [ [ [ [ [ [NN9MNNNNNQRR ww!!+...s)AAA)rrrr)r!N)r"r#r$r#r%r&)r3rr!)r4r#r5r-r6r-r$r#)r%rErFrE)NNT)rrvrwrvrxry)Nr!)rrrr#)r%rrr-rFr)__name__ __module__ __qualname____doc__r rr2rDrmrurrrrrrr __classcell__)rs@rrrsC<'4ko5555555 s s s s s     8@ @ @ @    "        rrr / / / / / / / / /r r)" __future__rrOrartypingrnumpyr>rZtorch.nnreultralytics.datarrultralytics.engine.trainerrultralytics.modelsr ultralytics.nn.tasksr ultralytics.utilsr r r ultralytics.utils.patchesrultralytics.utils.plottingrrultralytics.utils.torch_utilsrrrrNr rrsb#"""""  AAAAAAAA222222######//////7777777777666666????????TTTTTTTTP/P/P/P/P/{P/P/P/P/P/r