hht] hddlmZddlmZddlmZmZmZddlZddl Z ddl m Z m Z mZmZmZddlmZmZmZmZmZ erJeddusJddlZeed sJddlZdd lmZd d gZd ZdZdZdZ dZ!da"de%fdZ&de%fdZ'de(fdZ)de(fdZ*de+de+fdZ,de-fdZ.de-fdZ/deddfdZ0de1fdZ2dee+fdZ3d2dee1fdZ4d3dee1fd Z5d!e%d"edeeeefd#Z6deefd$Z7de1fd%Z8d4d&Z9d2d4d'Z:d4d(Z;d4d)Zd4d-Z?d4d.Z@d4d/ZAd4d0ZBere?e@eAeBd1ZCyiZCy#e#e$f$rdZYwxYw)5)Callable)SimpleNamespace)AnyListOptionalN)LOGGERRANKSETTINGS TESTS_RUNNINGops)ClassifyMetrics DetMetrics OBBMetrics PoseMetricsSegmentMetricscometT __version__)Pathdetectsegment)confusion_matrixconfusion_matrix_normalized)F1_curveP_curveR_curvePR_curve)labelslabels_correlogram)BoxMask)rPosereturnc`tjd}|tjd|Sy)zIReturn the Comet mode from environment variables, defaulting to 'online'. COMET_MODEaThe COMET_MODE environment variable is deprecated. Please use COMET_START_ONLINE to set the Comet experiment mode. To start an offline Comet experiment, use 'export COMET_START_ONLINE=0'. If COMET_START_ONLINE is not set or is set to '1', an online Comet experiment will be created.online)osgetenvrwarning) comet_modes _/var/www/html/dev/engine/venv/lib/python3.12/site-packages/ultralytics/utils/callbacks/comet.py_get_comet_moder+'s3<(J m  c.tjddS)zRReturn the Comet model name from environment variable or default to 'Ultralytics'.COMET_MODEL_NAME Ultralytics)r&r'r,r*_get_comet_model_namer16s 99' 77r,c@ttjddS)z[Get the evaluation batch logging interval from environment variable or use default value 1.!COMET_EVAL_BATCH_LOGGING_INTERVALintr&r'r0r,r* _get_eval_batch_logging_intervalr7;s ryyrNCOMET_START_ONLINEoffline10COMET_PROJECT_NAME) project_name)eval_batch_logging_intervallog_confusion_matrix_on_evallog_image_predictionsmax_image_predictionsz Created from ultralyticszEComet installed but not initialized correctly, not logging this run. )r r&r'r+environprojectcomet_mlstartlog_parametersvars log_othersr7rHrKr; log_other Exceptionrr()rLr) _project_name experimentes r*_resume_or_create_experimentrfUs 7 yy%&.$& 2< 2I3s '(d "6 E ^^? !!$t*-/O/Q0L0N)F)H)J)L    ^]; d^_`^abccdsBC D&DDcJ|jdz}t|jj|jz}||z}||j k(}|j j}|j j}||zdk(}|xr|dkDxr|xr| }t||||S)aH Return metadata for YOLO training including epoch and asset saving status. Args: trainer (ultralytics.engine.trainer.BaseTrainer): The YOLO trainer object containing training state and config. Returns: (dict): Dictionary containing current epoch, step, save assets flag, and final epoch flag. r4r) curr_epoch curr_step save_assets final_epoch) epochlen train_loaderdataset batch_sizeepochsrLsave save_perioddict) trainerrhtrain_num_steps_per_epochrirkrrrs save_intervalrjs r*_fetch_trainer_metadatarxys"J #G$8$8$@$@ AWEWEW W66I.K <<  D,,**K,1MP;?P}P[K : al mmr,c|\}}tj|||}tj||||}tj|}|ddxxx|dddz zccc|j }|S)a Scale bounding box from resized image coordinates to original image coordinates. YOLO resizes images during training and the label values are normalized based on this resized shape. This function rescales the bounding box labels to the original image shape. Args: box (torch.Tensor): Bounding box in normalized xywh format. resized_image_shape (tuple): Shape of the resized image (height, width). original_image_shape (tuple): Shape of the original image (height, width). ratio_pad (tuple): Ratio and padding information for scaling. Returns: (List[float]): Scaled bounding box coordinates in xywh format with top-left corner adjustment. )hwN)r xywhn2xyxy scale_boxes xyxy2xywhtolist)boxresized_image_shapeoriginal_image_shape ratio_padresized_image_heightresized_image_widths r*+_scale_bounding_box_to_original_image_shapersu$1D-- .. 48K LC //-s4H) TC -- CGs12w{G **,C Jr,c|d|k(}|d|}t|dk(rtjd|dy|d|jdj }|r|Dcgc]}t ||}}|d |}|d |} |d |} g} t ||D]5\} }t| | || } | j| gd |td d7d| dScc}w)a9 Format ground truth annotations for object detection. This function processes ground truth annotations from a batch of images for object detection tasks. It extracts bounding boxes, class labels, and other metadata for a specific image in the batch, and formats them for visualization or evaluation. Args: img_idx (int): Index of the image in the batch to process. image_path (str | Path): Path to the image file. batch (dict): Batch dictionary containing detection data with keys: - 'batch_idx': Tensor of batch indices - 'bboxes': Tensor of bounding boxes in normalized xywh format - 'cls': Tensor of class labels - 'ori_shape': Original image shapes - 'resized_shape': Resized image shapes - 'ratio_pad': Ratio and padding information class_name_map (dict, optional): Mapping from class indices to class names. Returns: (dict | None): Formatted ground truth annotations with the following structure: - 'boxes': List of box coordinates [x, y, width, height] - 'label': Label string with format "gt_{class_name}" - 'score': Confidence score (always 1.0, scaled by _scale_confidence_score) Returns None if no bounding boxes are found for the image. batch_idxbboxesr Comet Image: z has no bounding boxes labelsNclsr4 ori_shape resized_shapergt_g?boxeslabelr< ground_truthnamedata) rmrdebugsqueezerstrziprappendrA) img_idx image_pathbatchclass_name_mapindicesr cls_labelsrrrrrrs r*._format_ground_truth_annotations_for_detectionrs#6K G+G 8_W %F 6{a }ZL0MNOug&..q188:J>HIUc./0I I -g609k"7+I D&*- U9#?RThjst ug05   #D 11#Js$Cc,|j}|jr t|n|}|j|}|st j d|dy|r(|r&|j Dcic] \}}||| }}} ddlm} g} |D]k} | d} t| d} | d}|rt||}| g|| d }| '| jd d}| t|| }|||d <| j|md | d Scc}}w#t$rd} YwxYw)a Format YOLO predictions for object detection visualization. Args: image_path (Path): Path to the image file. metadata (dict): Prediction metadata containing bounding boxes and class information. class_label_map (dict, optional): Mapping from class indices to class names. class_map (dict, optional): Additional class mapping for label conversion. Returns: (dict | None): Formatted prediction annotations or None if no predictions exist. rz" has no bounding boxes predictionsNr)decodebboxr< category_idr segmentationpoints predictionr)stem isnumericr6getrritemsfaster_coco_eval.core.maskr ImportErrorrAr _extract_segmentation_annotationr)rmetadataclass_label_map class_maprimage_id predictionskvrrrrr< cls_labelannotation_datasegmentss r*_format_prediction_annotationsrsH ??D NN,s4y$H,,x(K  }ZL0RST97F7L7L7NOtq!9Q<?OO5 D!% 6"' 7(;<}- OI67I%*Gi%P  !~~nd;H#;HfM#,4) O$#%&!$ //7P s.C?D DDsegmentation_rawrc ||}tj|tjtj\}}|Dcgc]4}t |dk\st j |j6}}|Dcgc] }|jj"c}Scc}wcc}w#t$r"}tjd|Yd}~yd}~wwxYw)ag Extract segmentation annotation from compressed segmentations as list of polygons. Args: segmentation_raw (str): Raw segmentation data in compressed format. decode (Callable): Function to decode the compressed segmentation data. Returns: (List[List[Any]] | None): List of polygon points or None if extraction fails. z1Comet Failed to extract segmentation annotation: N) cv2 findContours RETR_LISTCHAIN_APPROX_SIMPLErmnparrayrravelrrbrr() rrmaskcontours_polygon annotations annotationres r*rrsP&'&&tS]]CIJ   "))+JJaJ PJ1#NOO Ps<AB5B+%B+=B5%B0(B5+ B55 C >CC czt||||}t||||}||fDcgc]}|| } }| r| gSdScc}w)aO Join the ground truth and prediction annotations if they exist. Args: img_idx (int): Index of the image in the batch. image_path (Path): Path to the image file. batch (dict): Batch data containing ground truth annotations. prediction_metadata_map (dict): Map of prediction metadata by image ID. class_label_map (dict): Mapping from class indices to class names. class_map (dict): Additional class mapping for label conversion. Returns: (List | None): List of annotation dictionaries or None if no annotations exist. N)rr) rrrprediction_metadata_maprrground_truth_annotationsprediction_annotationsrrs r*_fetch_annotationsr3sn" NUO <+_i '?@V%W![e[q K(K=1T1s88cpi}|D].}|j|dg||dj|0|S)zMCreate metadata map for model predictions by grouping them based on image ID.r) setdefaultr)model_predictionspred_metadata_maprs r*_create_prediction_metadata_maprQsN'E $$Z %;R@*Z0188DE r,c|jjj}t|jdj dgz}|j ||t|||y)z-Log the confusion matrix to Comet experiment.names background)matrixrmax_categoriesrlstepN) validatorrrlistrvalueslog_confusion_matrixrm)rdrurirhconf_matrs r*_log_confusion_matrixr[s_  1188H g&--/ 0L> AE##c%j Yb$r,c|r4t||D]$\}}|j||j||&y|D] }|j||j|"y)a Log images to the experiment with optional annotations. This function logs images to a Comet ML experiment, optionally including annotation data for visualization such as bounding boxes or segmentation masks. Args: experiment (comet_ml.Experiment): The Comet ML experiment to log images to. image_paths (List[Path]): List of paths to images that will be logged. curr_step (int): Current training step/iteration for tracking in the experiment timeline. annotations (List[List[dict]], optional): Nested list of annotation dictionaries for each image. Each annotation contains visualization data like bounding boxes, labels, and confidence scores. )rrr)rrN)r log_imager)rd image_pathsrirrrs r* _log_imagesrdsl&)+{&C k "J  *// _i j k& SJ  *// R Sr,c |jj}|tvry|j}|syt |}|j }|j }t|dd}t} t} t|D]j\} } | dz| zdk7r| d} t| D]F\}}t| k\ryt|}t||| |||}t||g||tdz a Hly)a Log predicted boxes for a single image during training. This function logs image predictions to a Comet ML experiment during model validation. It processes validation data and formats both ground truth and prediction annotations for visualization in the Comet dashboard. The function respects configured limits on the number of images to log. Args: experiment (comet_ml.Experiment): The Comet ML experiment to log to. validator (BaseValidator): The validator instance containing validation data and predictions. curr_step (int): The current training step for logging timeline. Notes: This function uses global state to track the number of logged predictions across calls. It only logs predictions for supported tasks defined in COMET_SUPPORTED_TASKS. The number of logged images is limited by the COMET_MAX_IMAGE_PREDICTIONS environment variable. Nrr4rim_file)r)r)rLtaskCOMET_SUPPORTED_TASKSjdictr dataloaderrgetattrr7r; enumerate_comet_image_prediction_countrrr)rdrrirrpredictions_metadata_maprrrbatch_logging_intervalrXrrrrrrs r*_log_image_predictionsr{s( >>  D (( OOE >uE%%JooO ;5I=?=?%j1/ 5 M3 3q 8 I& #,[#9 / GZ,0EEj)J,(# K  '   *Q . )' / /r,c \d}t|jjtr3tDcgc]!}t D]}|j ||dz #}}}nt|jjtr3tDcgc]!}tD]}|j ||dz #}}}nMt|jjttfr#tDcgc]}|j |dz }}| t||dtDcgc]}|j |dz }}t||dt|jjts1tDcgc]}|j |dz }}t||dyycc}}wcc}}wcc}wcc}wcc}w)a Log evaluation plots and label plots for the experiment. This function logs various evaluation plots and confusion matrices to the experiment tracking system. It handles different types of metrics (SegmentMetrics, PoseMetrics, DetMetrics, OBBMetrics) and logs the appropriate plots for each type. Args: experiment (comet_ml.Experiment): The Comet ML experiment to log plots to. trainer (ultralytics.engine.trainer.BaseTrainer): The trainer object containing validation metrics and save directory information. Examples: >>> from ultralytics.utils.callbacks.comet import _log_plots >>> _log_plots(experiment, trainer) Nz.pngz.jpg) isinstancermetricsrEVALUATION_PLOT_NAMESSEGMENT_METRICS_PLOT_PREFIXsave_dirrPOSE_METRICS_PLOT_PREFIXrrrCONFUSION_MATRIX_PLOT_NAMESr LABEL_PLOT_NAMES)rdruplot_filenamesplotsprefixconfusion_matrix_filenamesrlabel_plot_filenamess r* _log_plotsrs"N'##++^</ 5    &%5 5 5  G%%--{ ;/ 2    &%5 5 5  G%%-- J/G HI^_'**wd^;__!J5Ql!m'"2"2wd^"C!m!m 6= g''// AQabv 0 0fXT? BbbJ 4d; B)  ` "n cs&F&F#FF$)F)cht}|j|t|jddy)z'Log the best-trained model to Comet.ml.zbest.ptT)file_or_folder file_name overwriteN)r1 log_modelrbest)rdru model_names r* _log_modelrs,&(JC 4EQZfjkr,rict||jjd|t||jjd|y)z=Log samples of image batches for train, validation, and test.ztrain_batch*.jpgzval_batch*.jpgN)rrglob)rdruris r*_log_image_batchesr s> G,,112DEyQ G,,112BCYOr,c.t|jy)zRCreate or resume a CometML experiment at the start of a YOLO pre-training routine.N)rfrL)rus r*on_pretrain_routine_startr s .r,ctj}|syt|}|d}|d}|j|j |j d||y)z@Log metrics and save batch images at the end of training epochs.Nrhritrain)rrrl)r\get_running_experimentrx log_metricslabel_loss_itemstloss)rurdrrhris r*on_train_epoch_endrs_002J &w/H,'J%I733GMM'3RYbjtur,ctj}|syt|}|d}|d}|d}|j|j|||j|j |||dk(r ddlm}|j|||||syt||trt||||trt||j|yy) aM Log model assets at the end of each epoch during training. This function is called at the end of each training epoch to log metrics, learning rates, and model information to a Comet ML experiment. It also logs model assets, confusion matrices, and image predictions based on configuration settings. The function retrieves the current Comet ML experiment and logs various training metrics. If it's the first epoch, it also logs model information. On specified save intervals, it logs the model, confusion matrix (if enabled), and image predictions (if enabled). Args: trainer (BaseTrainer): The YOLO trainer object containing training state, metrics, and configuration. Examples: >>> # Inside a training loop >>> on_fit_epoch_end(trainer) # Log metrics and assets to Comet ML Nrhrirjrr4r)model_info_for_loggers)r\rrxrrlrultralytics.utils.torch_utilsrrrHrrKrr)rurdrrhrirjrs r*on_fit_epoch_endrs&002J &w/H,'J%I=)K7??*M7::IZHQH5g>YV`a z7##%j'9jI$&z7+<+r)s*%!&& HHjj  G  $$ $ 8] ++ + &y1#TH5"/,$%!  8s8 B#B >3> 5U SdS StS !d!d4!dHnn0 %[B42goptgu42n20bjkobp20jsHQYZ^_cdg_hZiQj*2 d^2<$S.</~) &?0,$    M ^$Hs6D%% D10D1