o W h'H@sddlZddlmZddlZddlZddlmZmZm Z ddl m Z ddl m Z mZddlmZddlmZmZmZddlmZmZGd d d e ZdS) N)Path)build_dataloaderbuild_yolo_dataset converter) BaseValidator)LOGGERops)check_requirements)ConfusionMatrix DetMetricsbox_iou)output_to_target plot_imagescseZdZdZd*fdd ZddZddZd d Zd d Zd dZ ddZ ddZ ddZ ddZ ddZddZd+ddZddZd d!Zd"d#Zd$d%Zd&d'Zd(d)ZZS),DetectionValidatora A class extending the BaseValidator class for validation based on a detection model. This class implements validation functionality specific to object detection tasks, including metrics calculation, prediction processing, and visualization of results. Attributes: nt_per_class (np.ndarray): Number of targets per class. nt_per_image (np.ndarray): Number of targets per image. is_coco (bool): Whether the dataset is COCO. is_lvis (bool): Whether the dataset is LVIS. class_map (list): Mapping from model class indices to dataset class indices. metrics (DetMetrics): Object detection metrics calculator. iouv (torch.Tensor): IoU thresholds for mAP calculation. niou (int): Number of IoU thresholds. lb (list): List for storing ground truth labels for hybrid saving. jdict (list): List for storing JSON detection results. stats (dict): Dictionary for storing statistics during validation. Examples: >>> from ultralytics.models.yolo.detect import DetectionValidator >>> args = dict(model="yolo11n.pt", data="coco8.yaml") >>> validator = DetectionValidator(args=args) >>> validator() Ncsht|||||d|_d|_d|_d|_d|_d|j_t |j d|_ t ddd|_|j|_dS)a Initialize detection validator with necessary variables and settings. Args: dataloader (torch.utils.data.DataLoader, optional): Dataloader to use for validation. save_dir (Path, optional): Directory to save results. pbar (Any, optional): Progress bar for displaying progress. args (dict, optional): Arguments for the validator. _callbacks (list, optional): List of callback functions. NFdetect)save_dirg?gffffff? )super__init__ nt_per_class nt_per_imageis_cocois_lvis class_mapargstaskr rmetricstorchlinspaceiouvnumelniou)self dataloaderrpbarr _callbacks __class__V/var/www/vscode/kcb/lib/python3.10/site-packages/ultralytics/models/yolo/detect/val.pyr,s zDetectionValidator.__init__cCsb|dj|jdd|d<|jjr|dn|dd|d<dD] }|||j||<q"|S)z Preprocess batch of images for YOLO validation. Args: batch (dict): Batch containing images and annotations. Returns: (dict): Preprocessed batch. imgT) non_blocking) batch_idxclsbboxes)todevicerhalffloat)r"batchkr(r(r) preprocessBs  (zDetectionValidator.preprocesscCs&|j|jjd}t|to#d|vo#|tjdp#|tjd|_ t|to1d|vo1|j |_ |j r:t n t tdt|jd|_|jj|jjoX|j sU|j oX|j O_|j|_t|j|_t|dd|_|j|j_|jj|j_t|j|jjd |_d |_g|_t gggggd |_!d S) z Initialize evaluation metrics for YOLO detection validation. Args: model (torch.nn.Module): Model to validate. cocoz val2017.txtztest-dev2017.txtlvisend2endF)ncconfr)tpr=pred_cls target_cls target_imgN)"datagetrsplit isinstancestrendswithosseprrrcoco80_to_coco91_classlistrangelennamesr save_jsonvaltrainingr<getattrr;rplotsplotr r=confusion_matrixseenjdictdictstats)r"modelrPr(r(r) init_metricsSs$ "((   zDetectionValidator.init_metricscCsddS)zBReturn a formatted string summarizing class metrics of YOLO model.z%22s%11s%11s%11s%11s%11s%11s)ClassImages InstanceszBox(PRmAP50z mAP50-95)r()r"r(r(r)get_descmszDetectionValidator.get_descc CsPtj||jj|jj|jjdkrdn|jd|jjp|jj|jj |j |jjdkd S)z Apply Non-maximum suppression to prediction outputs. Args: preds (torch.Tensor): Raw predictions from the model. Returns: (List[torch.Tensor]): Processed predictions after NMS. rrTobb)r< multi_labelagnosticmax_detr;rotated) rnon_max_suppressionrr=iourr< single_cls agnostic_nmsrer;)r"predsr(r(r) postprocessqs  zDetectionValidator.postprocessc Cs|d|k}|d|d}|d|}|d|}|djdd}|d |}t|rHt|tj||jd gd }tj||||d |||||d S)a Prepare a batch of images and annotations for validation. Args: si (int): Batch index. batch (dict): Batch data containing images and annotations. Returns: (dict): Prepared batch with processed annotations. r-r.r/ ori_shaper*N ratio_padr1)r:rr:rrp)r.bboxrnimgszrp) squeezeshaperMr xywh2xyxyrtensorr1 scale_boxes) r"sir4idxr.rsrnrtrpr(r(r)_prepare_batchs    "z!DetectionValidator._prepare_batchcCs:|}tj|d|ddddf|d|dd|S)a Prepare predictions for evaluation against ground truth. Args: pred (torch.Tensor): Model predictions. pbatch (dict): Prepared batch information. Returns: (torch.Tensor): Prepared predictions in native space. rtNrnrprr)clonerry)r"predpbatchprednr(r(r) _prepare_preds  $z DetectionValidator._prepare_predc Cst|D]\}}|jd7_t|}ttjd|jdtjd|jdtj||jtj|jdd}| ||}| d| d}} t|} ||d<| |d <|dkru| rt|j D] } |j | || qZ|jjrt|jjd | |d q|jjrd|d d d f<|||} | d d d f|d<| d d d f|d<| r|| | ||d<|jjr|j| | ||j D] } |j | || q|jjr|| |d||jjr|| |jj|d|jdt|d|jdqd S)z Update metrics with new predictions and ground truth. Args: preds (List[torch.Tensor]): List of predictions from the model. batch (dict): Batch data containing ground truth. r:rrq)dtyper1)r=r?r>r.rsr@rAN) detections gt_bboxesgt_clsr}r=r?r>im_filernlabelsz.txt) enumeraterVrMrXrzerosr1r!boolr|popuniquerYkeysappendrrSrU process_batchrir_process_batchrO pred_to_jsonsave_txt save_one_txt save_confrrstem) r"rkr4rzrnprstatrr.rsnlr5rr(r(r)update_metricssT    z!DetectionValidator.update_metricscOs|j|j_|j|j_dS)z Set final values for metrics speed and confusion matrix. Args: *args (Any): Variable length argument list. **kwargs (Any): Arbitrary keyword arguments. N)speedrrU)r"rkwargsr(r(r)finalize_metricss z#DetectionValidator.finalize_metricscCsdd|jD}tj|dt|jd|_tj|dt|jd|_| ddt |r>|j j di|d|j i|j jS) z Calculate and return metrics statistics. Returns: (dict): Dictionary containing metrics results. cSs&i|]\}}|t|dqS)r)rcatcpunumpy).0r5vr(r(r) s&z0DetectionValidator.get_stats..r@) minlengthrANon_plotr()rYitemsnpbincountastypeintr<rrrrMrprocessr results_dict)r"rYr(r(r) get_statss zDetectionValidator.get_statscCsddt|jj}t|d|j|jg|jR|jdkr0t d|j j d|j j rf|j sf|jdkrft|jrft|jjD]\}}t||j||j||j|g|j|RqG|j jr~dD]}|jj|j|j||jd qld Sd S) z0Print training/validation set metrics per class.z %22s%11i%11iz%11.3gallrzno labels found in z, set, can not compute metrics without labelsr:)TF)rrN normalizerN)rMrrrinforVrsum mean_resultswarningrrverboserQr<rYrap_class_indexrNr class_resultrSrUrTrvaluesr)r"pficrr(r(r) print_resultss (",z DetectionValidator.print_resultscCs4t||ddddf}||dddf||S)a] Return correct prediction matrix. Args: detections (torch.Tensor): Tensor of shape (N, 6) representing detections where each detection is (x1, y1, x2, y2, conf, class). gt_bboxes (torch.Tensor): Tensor of shape (M, 4) representing ground-truth bounding box coordinates. Each bounding box is of the format: (x1, y1, x2, y2). gt_cls (torch.Tensor): Tensor of shape (M,) representing target class indices. Returns: (torch.Tensor): Correct prediction matrix of shape (N, 10) for 10 IoU levels. Nr}r)r match_predictions)r"rrrrhr(r(r)rsz!DetectionValidator._process_batchrPcCst|j|||j||jdS)al Build YOLO Dataset. 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`. Returns: (Dataset): YOLO dataset. )modestride)rrrBr)r"img_pathrr4r(r(r) build_dataset$s z DetectionValidator.build_datasetcCs&|j||dd}t|||jjdddS)a Construct and return dataloader. Args: dataset_path (str): Path to the dataset. batch_size (int): Size of each batch. Returns: (torch.utils.data.DataLoader): Dataloader for validation. rP)r4rFrm)shufflerank)rrrworkers)r" dataset_path batch_sizedatasetr(r(r)get_dataloader2s z!DetectionValidator.get_dataloaderc CsHt|d|d|dd|d|d|jd|d|j|jd d S) z Plot validation image samples. Args: batch (dict): Batch containing images and annotations. ni (int): Batch index. r*r-r.rmr/r val_batchz _labels.jpgpathsfnamerNrN)rrurrNr)r"r4nir(r(r)plot_val_samples@s  z#DetectionValidator.plot_val_samplescCsFt|dgt||jjdR|d|jd|d|j|jddS)a Plot predicted bounding boxes on input images and save the result. Args: batch (dict): Batch containing images and annotations. preds (List[torch.Tensor]): List of predictions from the model. ni (int): Batch index. r*)rerrz _pred.jpgrN)rr rrerrNr)r"r4rkrr(r(r)plot_predictionsSs  z#DetectionValidator.plot_predictionscCsTddlm}|tj|d|dftjdd|j|ddddfdj||ddS) a Save YOLO detections to a txt file in normalized coordinates in a specific format. Args: predn (torch.Tensor): Predictions in the format (x1, y1, x2, y2, conf, class). save_conf (bool): Whether to save confidence scores. shape (tuple): Shape of the original image. file (Path): File path to save the detections. r)Resultsr:)rN)pathrNboxes)r)ultralytics.engine.resultsrrruint8rNr)r"rrrvfilerr(r(r)res  zDetectionValidator.save_one_txtc Cst|j}|r t|n|}t|ddddf}|ddddf|ddddfd8<t||D] \}}|j ||j t|ddd|Dt |dddq=dS)z Serialize YOLO predictions to COCO json format. Args: predn (torch.Tensor): Predictions in the format (x1, y1, x2, y2, conf, class). filename (str): Image filename. Nr}rorcSsg|]}t|dqS))roundrxr(r(r) sz3DetectionValidator.pred_to_json..)image_id category_idrsscore) rr isnumericrr xyxy2xywhziptolistrWrrr)r"rfilenamerrboxpbr(r(r)rxs 0  zDetectionValidator.pred_to_jsonc CsB|jjr|js |jrt|jr|jd}|jdd|jr"dnd|jjd}|jr0dnd}t d |d |d |d z||fD] }| sSJ|d qFt |jrZdnd|jrddl m}ddlm}|t|}|t|} ||| d} nddlm} m} | t|}|t|} | || d} dd|jjjD| j_| | | |jr| |jr| j ddn | j!d| j!dg\||j"j#d<||j"j#d<|jr| j!d|d<| j!d|d <| j!d!|d"<| j!d|d#<W|SW|St$y} zt %|d$| WYd} ~ |Sd} ~ ww|S)%a Evaluate YOLO output in JSON format and return performance statistics. Args: stats (dict): Current statistics dictionary. Returns: (dict): Updated statistics dictionary with COCO/LVIS evaluation results. zpredictions.jsonr annotationszinstances_val2017.jsonlvis_v1_z.json pycocotoolsr9z Evaluating z mAP using z and z...z file not foundzpycocotools>=2.0.6z lvis>=0.5.3r)COCO)COCOevalrs)LVISLVISEvalcSsg|] }tt|jqSr()rrrrr(r(r)rsz0DetectionValidator.eval_json..NroAPAP50rmAPrzmetrics/APr(B)APczmetrics/APc(B)APfzmetrics/APf(B)fitnessz unable to run: )&rrOrrrMrWrrBrDrris_filer pycocotools.cocorpycocotools.cocoevalrrFloadResr9rr _load_jsonr#rim_filesparamsimgIdsevaluate accumulate summarizerrYresultsrr Exceptionr)r"rY pred_json anno_jsonpkgrrrannorrPrrer(r(r) eval_jsons\$       &zDetectionValidator.eval_json)NNNNN)rPN)__name__ __module__ __qualname____doc__rr6r[rarlr|rrrrrrrrrrrrr __classcell__r(r(r&r)rs*7  r)rHpathlibrrrrultralytics.datarrrultralytics.engine.validatorrultralytics.utilsrrultralytics.utils.checksr ultralytics.utils.metricsr r r ultralytics.utils.plottingr rrr(r(r(r)s