o ߥi4@sddlmZddlZddlmZddlmZmZddl Z ddl m Z ddlmZddlmZddlmZmZddlmZddlmZmZdd lmZejeejd Gd d d eZd dZdS)N) OrderedDict)AnyDict)COCO)COCOeval)dumpload)Metrics)METRICSMetric) default_group) group_key module_namec@seZdZdZddZdeeefdeeeffddZdd Z dd d Z d dZ ddZ dddZ ddZddZddZddZddZdS) #ImageInstanceSegmentationCOCOMetriczMThe metric computation class for COCO-style image instance segmentation. cCsHd|_d|_ddg|_d|_tjddttdddd |_g|_ dS) Nbboxsegm)di,ig?gffffff?g!@T)endpoint) ann_fileclassesmetrics proposal_numsnplinspaceintroundiou_thrsresultsselfr!i/home/ubuntu/.local/lib/python3.10/site-packages/modelscope/metrics/image_instance_segmentation_metric.py__init__s  z,ImageInstanceSegmentationCOCOMetric.__init__outputsinputscCsf|d}t|dtrdd|D}|j||jdur1|ddd|_|ddd|_dSdS)N eval_resultrcSsg|] \}}|t|fqSr!)encode_mask_results).0 bbox_results mask_resultsr!r!r" &sz;ImageInstanceSegmentationCOCOMetric.add.. img_metasrr) isinstancetuplerextendrr)r r$r%resultr!r!r"add"s  z'ImageInstanceSegmentationCOCOMetric.addcCst|j}|j|jd|_||_||j|j\}}t }|j D]}|}||vr1t |dzt ||}|dkrF|D]}| dq>||} WntyZtdYnwt|| |} |j| j_|j| j_t|j| j_|j| j_dddd d d d d ddddd } | | | gd} | D]} |d| }t| j| | d}|||<q| jdd }|ddd|ddd|ddd|d dd|d dd|d d ||d<q"|dur||S)N)catNmsz is not in resultsrrz2The testing results of the whole dataset is empty.rr ) mAPmAP_50mAP_75mAP_smAP_mmAP_lzAR@100zAR@300zAR@1000z AR_s@1000z AR_m@1000z AR_l@1000)r=r>r?r@rArB_z.3f _mAP_copypaste) rr getCatIdsrcat_ids getImgIdsimg_idsformat_resultsrrrKeyErrorrpoploadRes IndexErrorprintrparamscatIdsimgIdslistrmaxDetsriouThrsevaluate accumulate summarizefloatstatscleanup)r cocoGt result_filestmp_dir eval_resultsmetriciou_type predictionsxcocoDtcocoEvalcoco_metric_names metric_items metric_itemkeyvalapr!r!r"rV-sr           0z,ImageInstanceSegmentationCOCOMetric.evaluateothercCs|j|jdSN)rr/)r rlr!r!r"mergessz)ImageInstanceSegmentationCOCOMetric.mergecCs|jSrm)rrr!r!r" __getstate__vsz0ImageInstanceSegmentationCOCOMetric.__getstate__cCs|||_dSrm)r#r)r stater!r!r" __setstate__ys z0ImageInstanceSegmentationCOCOMetric.__setstate__NcKsrt|ts Jdt|t|ksJdt|t||dur-t}t|jd}nd}| ||}||fS)aFormat the results to json (standard format for COCO evaluation). Args: results (list[tuple | numpy.ndarray]): Testing results of the dataset. data_infos(list[tuple | numpy.ndarray]): data information jsonfile_prefix (str | None): The prefix of json files. It includes the file path and the prefix of filename, e.g., "a/b/prefix". If not specified, a temp file will be created. Default: None. Returns: tuple: (result_files, tmp_dir), result_files is a dict containing the json filepaths, tmp_dir is the temporal directory created for saving json files when jsonfile_prefix is not specified. zresults must be a listz?The length of results is not equal to the dataset len: {} != {}Nr) r-rSlenformattempfileTemporaryDirectoryospjoinname results2json)r rrIjsonfile_prefixkwargsr^r]r!r!r"rJ}s z2ImageInstanceSegmentationCOCOMetric.format_resultscCs4|}|d|d|d|d|d|dgS)a.Convert ``xyxy`` style bounding boxes to ``xywh`` style for COCO evaluation. Args: bbox (numpy.ndarray): The bounding boxes, shape (4, ), in ``xyxy`` order. Returns: list[float]: The converted bounding boxes, in ``xywh`` order. rrr3r4)tolist)r r_bboxr!r!r" xyxy2xywhs  z-ImageInstanceSegmentationCOCOMetric.xyxy2xywhcCsg}tt|jD]8}|j|}||}t|jdD]%}t}||d<||||d<t||d|d<d|d<||qq |S)z,Convert proposal results to COCO json style.rimage_idrr5scorer category_id)rangerrrIshapedictr~rYappend)r r json_resultsidximg_idbboxesidatar!r!r"_proposal2jsons  z2ImageInstanceSegmentationCOCOMetric._proposal2jsonc Csg}tt|jD]Q}|j|}||}tt|D]?}|t|jkr%n5||}t|jdD](}t} || d<|||| d<t||d| d<|j|| d<| | q0qq |S)z-Convert detection results to COCO json style.rrrr5rr) rrrrIrrrr~rYrGr) r rrrrr0labelrrrr!r!r" _det2jsons$  z-ImageInstanceSegmentationCOCOMetric._det2jsoncCsg}g}tt|jD]}|j|}||\}}tt|D]}|t|jkr)n||} t| jdD](} t} || d<|| | | d<t| | d| d<|j|| d<| | q4t |t ro|d|} |d|} n ||} dd | D} t| jdD]A} t} || d<|| | | d<t| | | d<|j|| d<t | | d t r| | d  | | d <| | | d <| | qqq ||fS) z9Convert instance segmentation results to COCO json style.rrrr5rrrcSsg|]}|dqS)r5r!)r(rr!r!r"r+szBImageInstanceSegmentationCOCOMetric._segm2json..counts segmentation)rrrrIrrrr~rYrGrr-r.bytesdecode)r rbbox_json_resultssegm_json_resultsrrdetsegrrrrsegms mask_scorer!r!r" _segm2jsonsD        z.ImageInstanceSegmentationCOCOMetric._segm2jsoncCst}t|dtr&||}|d|d<|d|d<t||d|St|dtr[||}|d|d<|d|d<|d|d<t|d|dt|d|d|St|dtjrx| |}|d|d<t||d|St d ) a(Dump the detection results to a COCO style json file. There are 3 types of results: proposals, bbox predictions, mask predictions, and they have different data types. This method will automatically recognize the type, and dump them to json files. Args: results (list[list | tuple | ndarray]): Testing results of the dataset. outfile_prefix (str): The filename prefix of the json files. If the prefix is "somepath/xxx", the json files will be named "somepath/xxx.bbox.json", "somepath/xxx.segm.json", "somepath/xxx.proposal.json". Returns: dict[str: str]: Possible keys are "bbox", "segm", "proposal", and values are corresponding filenames. rz .bbox.jsonrproposalz .segm.jsonrrz.proposal.jsonzinvalid type of results) rr-rSrrr.rrndarrayr TypeError)r routfile_prefixr]rr!r!r"rys*   z0ImageInstanceSegmentationCOCOMetric.results2json)rlrrm)__name__ __module__ __qualname____doc__r#rstrrr1rVrnrorqrJr~rrrryr!r!r!r"rs"  F  ,rc Cst|tr |\}}n|}t|}ddt|D}tt|D]%}||D]}||ttj|ddddtj fddddq%qt|trN||fS|S)aEncode bitmap mask to RLE code. Args: mask_results (list | tuple[list]): bitmap mask results. In mask scoring rcnn, mask_results is a tuple of (segm_results, segm_cls_score). Returns: list | tuple: RLE encoded mask. cSsg|]}gqSr!r!)r(rCr!r!r"r+9sz'encode_mask_results..NFuint8)orderdtyper) r-r.rrrr mask_utilencoderarraynewaxis)r* cls_segmscls_mask_scores num_classesencoded_mask_resultsrcls_segmr!r!r"r')s*   r') os.pathpathrvrt collectionsrtypingrrnumpyrpycocotools.maskmaskrpycocotools.cocorpycocotools.cocoevalrmodelscope.fileiorrmodelscope.metainfor modelscope.metricsr r modelscope.utils.registryr register_moduleimage_ins_seg_coco_metricrr'r!r!r!r"s&