FijvddlmZddlZddlZddlmZddlmZddlm Z ddl m Z ddl m Z mZGdd e Zy) ) annotationsN)Image)classify_transforms) BasePredictor)Results) DEFAULT_CFGopscBeZdZdZeddfdfd ZfdZdZdZxZ S)ClassificationPredictoraA class extending the BasePredictor class for prediction based on a classification model. This predictor handles the specific requirements of classification models, including preprocessing images and postprocessing predictions to generate classification results. Attributes: args (dict): Configuration arguments for the predictor. Methods: preprocess: Convert input images to model-compatible format. postprocess: Process model predictions into Results objects. Examples: >>> from ultralytics.utils import ASSETS >>> from ultralytics.models.yolo.classify import ClassificationPredictor >>> args = dict(model="yolo26n-cls.pt", source=ASSETS) >>> predictor = ClassificationPredictor(overrides=args) >>> predictor.predict_cli() Notes: - Torchvision classification models can also be passed to the 'model' argument, i.e. model='resnet18'. NcJt||||d|j_y)asInitialize the ClassificationPredictor with the specified configuration and set task to 'classify'. This constructor initializes a ClassificationPredictor instance, which extends BasePredictor for classification tasks. It ensures the task is set to 'classify' regardless of input configuration. Args: cfg (dict): Default configuration dictionary containing prediction settings. overrides (dict, optional): Configuration overrides that take precedence over cfg. _callbacks (dict, optional): Dictionary of callback functions to be executed during prediction. classifyN)super__init__argstask)selfcfg overrides _callbacks __class__s j/var/www/pod-logistic/pod-ai/venv/lib/python3.12/site-packages/ultralytics/models/yolo/classify/predict.pyrz ClassificationPredictor.__init__'s! i4# c&t||t|jjdrt|jjjjddrM|jjjjdj t |jk7nd}|s|jjdk7rt|j|_y|jjj|_y)z9Set up source and inference mode and classify transforms. transformsrsizeFptN) r setup_sourcehasattrmodelrrmaximgszformatr)rsourceupdatedrs rrz$ClassificationPredictor.setup_source5s V$tzz''674::CSCSC^C^CiCijkClnt;u JJ   ' ' 2 21 5 : :c$**o M  07$**:K:Kt:S  + Y]YcYcYiYiYtYt rc &t|tjsitj|Dcgc]H}|j t j tj|tjJc}d}t|tjr|ntj|j|jj}|jjr|jS|j!Scc}w)zVConvert input images to model-compatible tensor format with appropriate normalization.r)dim) isinstancetorchTensorstackrr fromarraycv2cvtColor COLOR_BGR2RGB from_numpytordevicefp16halffloat)rimgims r preprocessz"ClassificationPredictor.preprocessAs#u||,++ade[]b#BSBS1T!UVeklC!ell3s9I9I#9NRRSWS]S]SdSde!ZZ__sxxz=#))+=fsA Dc Ft|tstj|ddddf}t|ttfr|dn|}t |||j dDcgc])\}}}t|||jj|+c}}}Scc}}}w)aProcess predictions to return Results objects with classification probabilities. Args: preds (torch.Tensor): Raw predictions from the model. img (torch.Tensor): Input images after preprocessing. orig_imgs (list[np.ndarray] | torch.Tensor): Original images before preprocessing. Returns: (list[Results]): List of Results objects containing classification results for each image. .Nr)pathnamesprobs) r'listr convert_torch2numpy_batchtuplezipbatchrrr;)rpredsr5 orig_imgspredorig_imgimg_paths r postprocessz#ClassificationPredictor.postprocessJs)T*55i@ddKI&utUm<a%-0y$**Q-,P  (h H84::3C3C4 P   s).B)rz dict | None) __name__ __module__ __qualname____doc__rrrr7rG __classcell__)rs@rr r s&.'$RV $  > rr ) __future__rr,r(PILrultralytics.data.augmentrultralytics.engine.predictorrultralytics.engine.resultsrultralytics.utilsrr r rrrTs-# 86..M mM r