o W h &@sddlmZddlZddlmZmZddlmZddlmZddl m Z ddl m Z m Z mZddlmZmZdd lmZmZmZGd d d eZdS) )copyN)ClassificationDatasetbuild_dataloader) BaseTrainer)yolo)ClassificationModel) DEFAULT_CFGLOGGERRANK) plot_images plot_results) is_parallelstrip_optimizertorch_distributed_zero_firstcseZdZdZeddffdd ZddZd!dd Zfd d Zd"d dZ d#ddZ ddZ ddZ ddZ d$ddZddZddZdd ZZS)%ClassificationTrainera A class extending the BaseTrainer class for training based on a classification model. This trainer handles the training process for image classification tasks, supporting both YOLO classification models and torchvision models. Attributes: model (ClassificationModel): The classification model to be trained. data (dict): Dictionary containing dataset information including class names and number of classes. loss_names (List[str]): Names of the loss functions used during training. validator (ClassificationValidator): Validator instance for model evaluation. Methods: set_model_attributes: Set the model's class names from the loaded dataset. get_model: Return a modified PyTorch model configured for training. setup_model: Load, create or download model for classification. build_dataset: Create a ClassificationDataset instance. get_dataloader: Return PyTorch DataLoader with transforms for image preprocessing. preprocess_batch: Preprocess a batch of images and classes. progress_string: Return a formatted string showing training progress. get_validator: Return an instance of ClassificationValidator. label_loss_items: Return a loss dict with labelled training loss items. plot_metrics: Plot metrics from a CSV file. final_eval: Evaluate trained model and save validation results. plot_training_samples: Plot training samples with their annotations. Examples: >>> from ultralytics.models.yolo.classify import ClassificationTrainer >>> args = dict(model="yolo11n-cls.pt", data="imagenet10", epochs=3) >>> trainer = ClassificationTrainer(overrides=args) >>> trainer.train() Ncs>|duri}d|d<|ddurd|d<t|||dS)a) Initialize a ClassificationTrainer object. This constructor sets up a trainer for image classification tasks, configuring the task type and default image size if not specified. Args: cfg (dict, optional): Default configuration dictionary containing training parameters. overrides (dict, optional): Dictionary of parameter overrides for the default configuration. _callbacks (list, optional): List of callback functions to be executed during training. Examples: >>> from ultralytics.models.yolo.classify import ClassificationTrainer >>> args = dict(model="yolo11n-cls.pt", data="imagenet10", epochs=3) >>> trainer = ClassificationTrainer(overrides=args) >>> trainer.train() Nclassifytaskimgsz)getsuper__init__)selfcfg overrides _callbacks __class__Z/var/www/vscode/kcb/lib/python3.10/site-packages/ultralytics/models/yolo/classify/train.pyr2s zClassificationTrainer.__init__cCs|jd|j_dS)z9Set the YOLO model's class names from the loaded dataset.namesN)datamodelr rrrrset_model_attributesKsz*ClassificationTrainer.set_model_attributesTcCst||jd|jd|otdkd}|r|||D]}|jjs,t|dr,|t |t j j r<|jj r<|jj |_q|D]}d|_qA|S)ad Return a modified PyTorch model configured for training YOLO. Args: cfg (Any): Model configuration. weights (Any): Pre-trained model weights. verbose (bool): Whether to display model information. Returns: (ClassificationModel): Configured PyTorch model for classification. ncchannels)r%chverbosereset_parametersT)rr!r loadmodulesargs pretrainedhasattrr* isinstancetorchnnDropoutdropoutp parameters requires_grad)rrweightsr)r"mr5rrr get_modelOs$     zClassificationTrainer.get_modelcsdddl}t|j|jjvr!|jj|j|jjrdndd|_d}nt}t |j|j d|S)z Load, create or download model for classification tasks. Returns: (Any): Model checkpoint if applicable, otherwise None. rN IMAGENET1K_V1)r8r%) torchvisionstrr"models__dict__r-r.r setup_modelrreshape_outputsr!)rr<ckptrrrr@hs  z!ClassificationTrainer.setup_modeltraincCst||j|dk|dS)a Create a ClassificationDataset instance given an image path and mode. Args: img_path (str): Path to the dataset images. mode (str): Dataset mode ('train', 'val', or 'test'). batch (Any): Batch information (unused in this implementation). Returns: (ClassificationDataset): Dataset for the specified mode. rC)rootr-augmentprefix)rr-)rimg_pathmodebatchrrr build_dataset{s z#ClassificationTrainer.build_datasetrcCs|t||||}Wdn1swYt|||jj|d}|dkrReturns an instance of ClassificationValidator for validation.loss)r-r) rbrrClassificationValidator test_loadersave_dirrr- callbacksr#rrr get_validatorsz#ClassificationTrainer.get_validatorcs>fdd|jD}|dur|Stt|dg}tt||S)a_ Return a loss dict with labelled training loss items tensor. Args: loss_items (torch.Tensor, optional): Loss tensor items. prefix (str): Prefix to prepend to loss names. Returns: (Dict[str, float] | List[str]): Dictionary of loss items or list of loss keys if loss_items is None. csg|] }d|qS)/r).0xrFrr sz:ClassificationTrainer.label_loss_items..N)rbroundfloatdictzip)r loss_itemsrFkeysrrmrlabel_loss_itemss  z&ClassificationTrainer.label_loss_itemscCst|jd|jddS)zPlot metrics from a CSV file.T)fileron_plotN)r csvrxr#rrr plot_metricssz"ClassificationTrainer.plot_metricscCs|j|jfD]9}|r?t|||jur?td|d|jj|jj_|jj |jj_ |j|d|_ |j dd| dqdS)z3Evaluate trained model and save validation results.z Validating z...)r"fitnessNon_fit_epoch_end) lastbestexistsrr infor-r! validatorplotsmetricspop run_callbacks)rfrrr final_evals  z ClassificationTrainer.final_evalcCsBt|dtt|d|dd|jd|d|jddS)z Plot training samples with their annotations. Args: batch (Dict[str, torch.Tensor]): Batch containing images and class labels. ni (int): Number of iterations. rVrWr' train_batchz.jpg)images batch_idxrWfnamerxN)r r1arangeraviewrgrx)rrInirrrplot_training_sampless  z+ClassificationTrainer.plot_training_samples)NNT)rCN)rKrrC)NrC)__name__ __module__ __qualname____doc__rrr$r:r@rJrUrZrcrirvrzrr __classcell__rrrrrs!      r)rr1ultralytics.datarrultralytics.engine.trainerrultralytics.modelsrultralytics.nn.tasksrultralytics.utilsrr r ultralytics.utils.plottingr r ultralytics.utils.torch_utilsr rrrrrrrs