Datasets
Models
Downloads: 1
171 lines (167 with data), 21.5 kB
B
¸Yf ã @ sP d Z ddlZddlZddlZddlZddlZddlmZ ddlZ ddl
Z
ddl m Z ddl Z ddl
ZddlmZ ee ¡ Zejd Zeeejkr¤ej ee¡ eej ee ¡ ¡ZddlmZ ddl m Z ddl m Z dd l m!Z!m"Z"m#Z#m$Z$m%Z%m&Z&m'Z'm(Z(m)Z)m*Z*m+Z+m,Z,m-Z-m.Z.m/Z/m0Z0m1Z1m2Z2m3Z3 dd
l4m5Z5m6Z6m7Z7 dd l8m9Z9m:Z:m;Z; dd l<m=Z=m>Z> dd
l?m@Z@ ddlAmB mCZD ddlEZEddlFmGZG ddlHmIZImJZJmKZKmLZL dd ZMdd ZNdd ZOdd ZPe> dddddd d d d d d d d d d d ed! d"d d#d dded d#e dfd$d%ZQd&d' ZRd(d) ZSeTd*krLeR ZUeSeU dS )+aç
Validate a trained YOLOv5 detection model on a detection dataset
Usage:
$ python val.py --weights yolov5s.pt --data coco128.yaml --img 640
Usage - formats:
$ python val.py --weights yolov5s.pt # PyTorch
yolov5s.torchscript # TorchScript
yolov5s.onnx # ONNX Runtime or OpenCV DNN with --dnn
yolov5s_openvino_model # OpenVINO
yolov5s.engine # TensorRT
yolov5s.mlmodel # CoreML (macOS-only)
yolov5s_saved_model # TensorFlow SavedModel
yolov5s.pb # TensorFlow GraphDef
yolov5s.tflite # TensorFlow Lite
yolov5s_edgetpu.tflite # TensorFlow Edge TPU
yolov5s_paddle_model # PaddlePaddle
é N)ÚPath)Útqdm)Útabulate)ÚDetectMultiBackend)Ú Callbacks)Úcreate_dataloader)ÚLOGGERÚTQDM_BAR_FORMATÚProfileÚ
check_datasetÚcheck_img_sizeÚcheck_requirementsÚ
check_yamlÚcoco80_to_coco91_classÚcolorstrÚincrement_pathÚnon_max_suppressionÚ
print_argsÚ scale_boxesÚ xywh2xyxyÚextract_roi_featuresÚ xyxy2xywhÚ get_object_level_feature_maps2Ú xywh_to_xyxyÚget_fixed_xyxy)ÚConfusionMatrixÚ ap_per_classÚbox_iou)Úoutput_to_targetÚ plot_imagesÚplot_val_study)Ú
select_deviceÚsmart_inference_mode)ÚMyCNN)Ú roi_align)Úaccuracy_scoreÚprecision_scoreÚ recall_scoreÚf1_scorec
C s¨ t |¡ddddg }x| ¡ D ]^ }}}tt |¡ dd¡| d¡ ¡ }|rb|f||fn|f|} t|d$}
|
dt| ¡ | d ¡ W d Q R X q W d S )Né r é éÿÿÿÿÚaz%g Ú
) ÚtorchÚtensorÚtolistr ÚviewÚopenÚwriteÚlenÚrstrip) ÚprednÚ save_confÚshapeÚfileÚgnÚxyxyÚconfÚclsÚxywhÚlineÚf© rA úB/home/iml/Desktop/Talha/yolov5_P2_ROI_HCM_Micro_1000_Github/val.pyÚ save_one_txt? s $ rC c C sº |j ¡ rt|j n|j }t| d d
d d
f }|d d
d d
f |d d
dd
f d 8 < xRt| ¡ | ¡ D ]<\}}| ||t|d dd |D t|d dd¡ qvW d S )Nr* é é c S s g | ]}t |d qS )é )Úround)Ú.0ÚxrA rA rB ú
<listcomp>R s z!save_one_json.<locals>.<listcomp>)Úimage_idÚ category_idÚbboxÚscore)ÚstemÚ isnumericÚintr Úzipr0 ÚappendrG )r6 ÚjdictÚpathÚ class_maprK ÚboxÚpÚbrA rA rB Ú
save_one_jsonI s 0 rZ c C s t | jd |jd f¡ t¡}t|dd
dd
f | dd
dd
f }|dd
dd
f | dd
df k}x tt|D ]ü}t ||| k|@ ¡}|d jd r~t
t |d¡||d |d f dd
df fd¡ ¡
¡ }|d jd dkr\||dd
df ¡ ddd
}|t j|dd
df dd d }|t j|dd
df dd d }d||dd
df t¡|f< q~W tj|tj|jd
S ) zì
Return correct prediction matrix.
Arguments:
detections (array[N, 6]), x1, y1, x2, y2, conf, class
labels (array[M, 5]), class, x1, y1, x2, y2
Returns:
correct (array[N, 10]), for 10 IoU levels
r Nr) r* rE rD r+ T)Ú return_index)ÚdtypeÚdevice)ÚnpÚzerosr8 ÚastypeÚboolr Úranger4 r. ÚwhereÚcatÚstackÚcpuÚnumpyÚargsortÚuniquerQ r/ r] ) Ú
detectionsÚlabelsÚiouvÚcorrectÚiouÚ
correct_classÚirI ÚmatchesrA rA rB Ú
process_batchU s
*$<""""rr c
C s¤ t | jd |jd f¡ t¡}t|dd
dd
f | dd
dd
f }|dd
dd
f | dd
df k}tj|dd}x tt |D ]ü}t
||| k|@ ¡}|d jd rt t |d¡||d |d f dd
df fd¡
¡ ¡ } |d jd dkrj| | dd
df ¡ ddd
} | t j| dd
df d d
d } | t j| dd
df d d
d } d || dd
df t¡|f< qW tj|tj|jd |fS ) ax
Return correct prediction matrix and top indices.
Arguments:
detections (array[N, 6]), x1, y1, x2, y2, conf, class
labels (array[M, 5]), class, x1, y1, x2, y2
iouv (tensor[10]), IoU thresholds
Returns:
correct (array[N, 10]), for 10 IoU levels
top_indices (tensor[M]), top IoU-gaining detection indices for each label
r Nr) r* rE )ÚdimrD r+ T)r[ )r\ r] )r^ r_ r8 r` ra r r. Úargmaxrb r4 rc rd re rf rg rh ri rQ r/ r] )
rj rk rl rm rn ro Ú top_indicesrp rI rq rA rA rB Úmy_process_batcho s *$<""""rv é i gü©ñÒMbP?g333333ã?i, ÚvalÚ é Fzruns/valÚexpTc ¶ s, |d k } | rNt | ¡ jdddf\} } } } || jdkM }|rD| ¡ n| ¡ n¸t| |d} tt|| |d}|rz|d n|j ddd t
|| || |d}|j |j |j
|jf\}!} } } t||!d }|j}| rÔ|j}n*|j} | sþ| sþd
}t d | d | d
¡ t| } | ¡ | jdk}"t| d¡to@| d dtj d¡}#| rLd
n
t | d }$t j ddd| d}%|% ¡ }&| s&| r°| s°|j j!}'|'|$ks°t"| d|' d|$ d|j#| r¾d
n|d||fd |dkrÜd nd| f\}(})|d krö|nd}t$| | |||!| |(|)|
t%| d d d }d t&|$d!}*t'|d"rF|j(n|j)j(}+t|+t*t+frjt,t-|+}+|#rvt. n
t*t/d#},d$d% }-d&\
}.}/}0}1}2}3}4}5}6}7t0| dt0| dt0| df}8t j1d| d}9g g g g f\}:};}<}=| 2d'¡ t3||-t4d(}>g }?xæt-|>D ]Ø\}@\}A}B}Cd }D| 2d)¡ |8d P |"rP|Aj5| dd*}A 5| ¡|r^|A ¡ n|A ¡ }A|Ad+ }A|Aj6\}E}F}G}HW d Q R X |8d
, | r||An||A| d,d f\\}I}J}KW d Q R X | rè|9| |Jd d
d d-
f d
7 }9d d
d.d/
f }Ld d
d d-
f d d
d0d
f t j7|H|G|H|Gf| d9 < |rZfd1d2t/|ED ng }M|8d0 t8|I|||Md| |d3}IW d Q R X xòt-|ID ]ä\}N}Od d
d f |Nkd
d
f }Pt9|Od krtt j:|Nt ;|Oj6d d
¡ 5| ¡ |Od d
d d4
f 5| ¡gd
d5}Q|Kd |N 5t j<¡ =d ¡}R|Kd
|N 5t j<¡ =d ¡}St j> ?¡ g }Tx"t/t9|OD ]}U|O|Ud d4
f }Vt 7|Aj6d0 |Aj6d |Aj6d0 |Aj6d g¡ 5| ¡}W|V|W }Xt 7|Kd |N j6d
|Kd |N j6d0 |Kd |N j6d
|Kd |N j6d0 g¡ 5| ¡}Yt 7|Kd
|N j6d
|Kd
|N j6d0 |Kd
|N j6d
|Kd
|N j6d0 g¡ 5| ¡}Z|X|Y }[|X|Z }\t@|[|R\}]}^}_}`t@|\|S\}a}b}c}dt j7|]|^|_|`g| d ¡ }et j7|a|b|c|dg| d ¡ }ft j7d gt j<| d6}gt :|g|eg¡}ht :|g|fg¡}itA|R|h =d ¡ 5| ¡d7d8}jtA|S|i =d ¡ 5| ¡d7d8}kt j:|j|kgd
d5}l| ¡ 5| ¡ ||l}mtBjC|m Dd-d0¡d
d5}nt jE|nd
d5}o|T F|o¡ q^W |Pj6d |Oj6d }p}qt|B|N |C|N d }r}st j1|q|&t jG| d6}t d
7 |qd kr |pr|; F|tft j1d9| d|Pd d
d f f¡ |r|*jHd |Pd d
d f d: q| r4d |Od d
d;f< |O I¡ }utJ|A|N j6d
d
|ud d
d d4
f |s|C|N d
|p
rÞtK|Pd d
d
d;
f }vtJ|A|N j6d
d
|v|s|C|N d
t :|Pd d
d d
f |vfd
¡}wtH|u|w|%}ttL|u|w|%\}F}x|Q|x }yg }zd }Dx4t/t9|xD ]$}U|T|x|U M¡ N¡ }{|z F|{¡ q W t|r}|||jO}}g }~tPd< }|d= }|j ddd | Q¡ r¦| Rd>¡&}| S¡ }t |d }Dt|d
}W d Q R X n4d }Dd?}| Rd@¡ }| T|D dA| dA¡ W d Q R X dBdCdDdEdFdGg}dHd2 t/d-D }
dId2 t/d-D }xt/t9|zD ]}Ud }|L|U M¡ N¡ ¡ }tUdJdK |D
r |~ F|}¡ xBt/d-D ]6}|
| F|| V¡ ¡ || F|z|U | V¡ ¡
q`W
q W tW|~f|
| }|? X|¡ |
rÞtL|u|w|%dL\}}|* H|u|w¡ |; F|t|Od d
d4f |Od d
d;f |Pd d
d f f¡ | rL|d j ddd tY|u||s|d |rjZ dM dN | r`t[|u|:|r|, | 2dO|O|u|r|+|A|N ¡ qW | rÊ|@dk rÊt\|A|B|dP|@ dQ |+ t\|At]|I|B|dP|@ dR |+ | 2dS|@|A|B|C|I¡ qW tPd< }|d= }dBdCdDdEdFdGg}|j ddd | Q¡ rV| Rd>¡&}| S¡ }t |d }Dt|d
}W d Q R X nd }Dd?}|| dT|D dU }tR|dVdWdXX}t^ _|¡}tj` a|¡d k rÊdYgdZd2 |D d[d2 |D }| b|¡ | c|?¡ W d Q R X tdje|d dYdBdCdDdEdFdGd\d]d^d_d`dag
db}|jfd
d
}tdjg|dB dcdd}tdjg|dC dcdd}tdjg|dD dcdd}tdjg|dE dcdd}tdjg|dF dcdd}tdjg|dG dcdd}tdjg|d\ dcdd}tdjg|d] dcdd}tdjg|d^ dcdd}tdjg|d_ dcdd}tdjg|d` dcdd}tdjg|da dcdd}|th i|¡ jt ¡}|th i|¡ jt ¡}|th i|¡ jt ¡}|th i|¡ jt ¡}|th i|¡ jt ¡}|th i|¡ jt ¡}|th i|¡ jt ¡}|th i|¡ jt ¡}|th i|¡ jt ¡}|th i|¡ jt ¡}|th i|¡ jt ¡}|th i|¡ jt ¡}dedf }dgd2 tW|; D };t9|;r|;d k¡ rtl|;|||+dh\}.}/}0}1}2}<}=|<d d
d f |< md
¡ }6}<|0 m¡ |1 m¡ |6 m¡ |< m¡ f\}3}4}5}7thjn|;d jt ¡|$di} dj}¡t |¡dk | o¡ |3|4|5|7f ¡ | o¡ d krät pdl| dm¡ |
sú|$dnk rZ| sZ|$d
krZt9|;rZxJt-|=D ]>\}U}¢t |¡|+|¢ | |¢ |0|U |1|U |6|U |<|U f ¡ qW g }£|do||}¤|dp||}¥|dD||}¦|dE||}§|dq||}¨|dr||}©th m|¤ds |¥ds |¦ds |§ds |¨ds |©ds g¡}ª|£ X|¤|¥|¦|§|¨|©g¡ dtdudvdwdxg}«tqtr|£|«dydz |ª|krH|ª}| s¡ 5| ¡ t t| u¡ |d{| dT|D d| ¡ |Dd
7 }D| Rd@¡}| T|D dA| ¡ W d Q R X | s¡ 5| ¡ t t| u¡ |d} ¡ t+ fd~dK|8D }¬| sØ|d||f}st d|s |¬ ¡ |r|*jv|t*|+ w¡ d | 2d| |.|/|0|1|2|<|6|=|*¡ |rt9|:r|d k rJtt|t*rB|d n|jZndW}ttd}®tj` Q|®¡s|tj` x| d dd
¡}®t|| d }¯t d|¯ d¡ tR|¯d@}ty z|:|¡ W d Q R X yt{d d dl|m}}° d dl~m}± |°|®}²|² |¯¡}O|±|²|Od}³|#r dd2 |jjD |³j_|³
¡ |³ ¡ |³ ¡ |³jd d0
\}7}5W n4 tk
r~ }´ zt d|´ ¡ W d d }´~´X Y nX | ¡ | sÔ|r¶dAt9t*| d¡ d|d ndW}-t dt%d| |- ¡ th 1|$¡|7 }µx"t-|=D ]\}U}¢|<|U |µ|¢< qìW |3|4|5|7f|9 N¡ t9| ¡ |µ|¬fS )NTFrf )Ú
batch_size)Úexist_okrk )Úparentsr} )r] ÚdnnÚdataÚfp16)Úsr) z-Forcing --batch-size 1 square inference (1,3,ú,z) for non-PyTorch modelsrx Úcocoz val2017.txtÚncg à?gffffffî?é
)r] z (z< classes) trained on different --data than what you passed (zV classes). Pass correct combination of --weights and --data that are trained together.rF )ÚimgszÚspeed)g F)Útrainrx Útestz: )ÚpadÚrectÚworkersÚprefixr )r
Únamesiè z %22s%11s%11s%11s%11s%11s%11s)ÚClassÚImagesÚ InstancesÚPÚRÚmAP50zmAP50-95)
g g g g g g g g g g Ú on_val_start)ÚdescÚ
bar_formatÚon_val_batch_start)Ú non_blockingéÿ )Úaugmenté é é
rD c s, g | ]$} d d
df |kdd
f qS )Nr r) rA )rH rp )ÚtargetsrA rB rJ s zrun.<locals>.<listcomp>)rk Ú multi_labelÚagnosticÚmax_detr* )rs )r\ r] )é é )Ú output_size)rD r )rj rk rE ZAttribute_modelzcheckpoint.txtÚrg Úwr- zNuclear Chromatinz
Nuclear ShapeZNucleusZ CytoplasmzCytoplasmic BasophiliazCytoplasmic Vacuolesc S s g | ]}g qS rA rA )rH Ú_rA rA rB rJ s c S s g | ]}g qS rA rA )rH r© rA rA rB rJ s c s s | ]}|d kV qdS )rD NrA )rH rI rA rA rB ú <genexpr> s zrun.<locals>.<genexpr>)rl z.txt)r9 Zon_val_image_endÚ val_batchz _labels.jpgz _pred.jpgÚon_val_batch_endr© z.csvr, ry )ÚnewlineÚfilenamec S s g | ]
}| qS rA rA )rH ÚattrrA rA rB rJ Ô s c S s g | ]}d | qS )Zpred_rA )rH r¯ rA rA rB rJ Ô s zpred_Nuclear Chromatinzpred_Nuclear ShapeZ pred_NucleusZpred_Cytoplasmzpred_Cytoplasmic Basophiliazpred_Cytoplasmic Vacuoles)Úheaderr Úcoerce)Úerrorsc S sb |t |¡ t¡}|t |¡ t¡}t||}t||}t||}t||}| ||||gS )N)r^ Úisnanr` rQ r% r& r' r( )Úattribute_nameZ true_labelsZ pred_labelsÚaccuracyÚ precisionÚrecallÚf1rA rA rB Úcompute_and_print_metrics s
z&run.<locals>.compute_and_print_metricsc S s g | ]}t |d ¡ ¡ ¡ qS )r )r. rd rf rg )rH rI rA rA rB rJ s )ÚplotÚsave_dirr )Ú minlengthz$%22s%11i%11i%11.3g%11.3g%11.3g%11.3gÚallu" WARNING â ï¸ no labels found in z, set, can not compute metrics without labelsé2 ZNuclear_ChromatinZ
Nuclear_ShapeZCytoplasmic_BasophiliaZCytoplasmic_Vacuolesr+ Ú AttributeZAccuracyÚ PrecisionÚRecallÚF1Úgrid)ÚheadersÚtablefmtZ
best_weights_z.pthzlast_weights.pthc 3 s | ]}|j d V qdS )g @@N)Út)rH rI )ÚseenrA rB rª L s zKSpeed: %.1fms pre-process, %.1fms inference, %.1fms NMS per image at shape )r» r Ú
on_val_endz3../datasets/coco/annotations/instances_val2017.jsonrU Ú annotationszinstances_val2017.jsonz_predictions.jsonz&
Evaluating pycocotools mAP... saving z...zpycocotools>=2.0.6)ÚCOCO)ÚCOCOevalrM c S s g | ]}t t|jqS rA )rQ r rO )rH rI rA rA rB rJ j s zpycocotools unable to run: z labels/*.txtz labels saved to zResults saved to Úbold)ÚnextÚ
parametersr] ÚtypeÚhalfÚfloatr! r r Úmkdirr ÚstrideÚptÚjitÚenginer r r| r Úinfor ÚevalÚ
isinstanceÚgetÚstrÚendswithÚosÚseprQ r. ÚlinspaceÚnumelÚmodelr
ÚAssertionErrorÚwarmupr r r Úhasattrr ÚmoduleÚlistÚtupleÚdictÚ enumerater rb r
r_ Úrunr r Útor8 r/ r r4 rd ÚonesÚfloat32Ú unsqueezeÚcudaÚ empty_cacher r$ ÚFÚsoftmaxr1 rt rS ra rr Úcloner r rv Údetachrf ÚnameÚROOTÚexistsr2 Ú readlinesr3 r½ ÚitemrR ÚextendrC rO rZ r r ÚcsvÚwriterrU ÚgetsizeÚwriterowÚ writerowsÚpdÚread_csvÚilocÚ
to_numericr^ r³ r` Úanyr ÚmeanÚbincountÚsumÚwarningÚprintr r ÚsaveÚ
state_dictrº ÚvaluesÚjoinÚjsonÚdumpr
Úpycocotools.cocorÊ Úpycocotools.cocoevalrË ÚloadResÚdatasetÚim_filesÚparamsÚimgIdsÚevaluateÚ
accumulateÚ summarizeÚstatsÚ ExceptionÚglobr0 )¶r Z
cell_modelÚweightsr| r Ú
conf_thresÚ iou_thresr£ Útaskr] r Ú
single_clsr ÚverboseÚsave_txtÚ save_hybridr7 Ú save_jsonÚprojectrõ r} rÐ r rá Ú
dataloaderr» ÚplotsÚ callbacksÚ compute_lossÚtrainingrÔ rÕ rÖ rÓ rï Úis_cocor
rl ÚniouZncmr r Úconfusion_matrixr rV r ÚtpÚfprX r§ r¸ ÚmpÚmrÚmap50Úap50ÚmapÚdtÚlossrT r ÚapZap_classÚpbarZall_rowsÚbatch_iÚimÚpathsÚshapesÚepochÚnbr© ÚheightÚwidthÚpredsZ train_outZ int_featsZattribute_targetsÚlbÚsiÚpredrk ÚboxesZ int_feats_p2Z int_feats_p3Zall_top_indices_cell_predrp Z pred_tensorZimg_shape_tensorÚnormalized_xyxyÚp2_feature_shape_tensorÚp3_feature_shape_tensorÚp2_normalized_xyxyÚp3_normalized_xyxyÚp2_x_minÚp2_y_minÚp2_x_maxÚp2_y_maxÚp3_x_minÚp3_y_minÚp3_x_maxÚp3_y_maxÚp2_roiÚp3_roiÚ batch_indexÚp2_roi_with_batch_indexÚp3_roi_with_batch_indexÚp2_resized_objectÚp3_resized_objectÚ
concat_boxZoutput_cell_predictionZoutput_cell_prediction_probZtop_indices_cell_predÚnlÚnprrU r8 rm r6 ÚtboxÚlabelsnZmax_iou_indicesZmax_iou_predicted_boxesZattributes_predictionZattributes_predZ path_objectÚ file_nameÚ filenamesZcheckpoint_dirZcheckpoint_file_pathr@ ÚlinesZ
best_accuracyZattribute_namesZall_labels_listZ all_prediction_attributes_listÚcountÚlabelÚjÚrowsZcnf_ÚindicesZattribute_model_dirZcsv_file_path_with_epochÚcsvfileZ
csv_writerr° ÚdfZNuclear_Chromatin_arrayZNuclear_Shape_arrayZ
Nucleus_arrayZCytoplasm_arrayZ Cytoplasmic_Basophilia_arrayZCytoplasmic_Vacuoles_arrayZ pred_Nuclear_Chromatin_arrayZpred_Nuclear_Shape_arrayZpred_Nucleus_arrayZpred_Cytoplasm_arrayZ!pred_Cytoplasmic_Basophilia_arrayZ pred_Cytoplasmic_Vacuoles_arrayr¹ ÚntÚpfÚcÚmetricsZNC_precZNS_precZN_precZC_precZCB_precZCV_precZaverage_precisionrÄ rÆ r¨ Ú anno_jsonÚ pred_jsonrÊ rË ÚannorØ ÚeÚmapsrA )rÇ r rB rê sR !
* $
0".
@
0PP
*4$
&
6
"
* $ *: .
" ,
",rê c C sÔ t ¡ } | jdttd dd | jddttd dd | jd
td d d | jd
ddtddd | jdtddd | jdtddd | jdtddd | jdd d d | jd d d!d | jd"td#d$d | jd%d&d'd( | jd)d&d*d( | jd+d&d,d( | jd-d&d.d( | jd/d&d0d( | jd1d&d2d( | jd3d&d4d( | jd5td6 d7d | jd8d9d7d | jd:d&d;d( | jd<d&d=d( | jd>d&d?d( | ¡ }t|j |_ | j
|j d@¡O _
| j |j
O _ tt| |S )ANz--datazdata/coco128.yamlzdataset.yaml path)rÏ ÚdefaultÚhelpz --weightsú+z
yolov5s.ptz
model path(s))ÚnargsrÏ rs rt z --batch-sizerw z
batch sizez--imgszz--imgz
--img-sizei zinference size (pixels)z --conf-thresgü©ñÒMbP?zconfidence thresholdz --iou-thresg333333ã?zNMS IoU thresholdz --max-deti, z maximum detections per imagez--taskrx z train, val, test, speed or study)rs rt z--devicery z%cuda device, i.e. 0 or 0,1,2,3 or cpuz --workersrz z-max dataloader workers (per RANK in DDP mode)z --single-clsÚ
store_truez treat as single-class dataset)Úactionrt z --augmentzaugmented inferencez --verbosezreport mAP by classz
--save-txtzsave results to *.txtz
--save-hybridz-save label+prediction hybrid results to *.txtz --save-confz%save confidences in --save-txt labelsz --save-jsonz save a COCO-JSON results filez --projectzruns/valzsave to project/namez--namer{ z
--exist-okz*existing project/name ok, do not incrementz--halfz!use FP16 half-precision inferencez--dnnz!use OpenCV DNN for ONNX inferencez coco.yaml)ÚargparseÚArgumentParserÚ add_argumentrÛ rö rQ rÑ Ú
parse_argsr r r% rÜ r# r$ r Úvars)ÚparserÚoptrA rA rB Ú parse_opt} s: r c C sº t td dd | jdkrZ| jdkr8t d| j d¡ | jrHt d¡ tf t| n\t | j
t rl| j
n| j
g}t j
¡ o| jd k| _| jd
krÌd \| _| _| _x$|D ] | _
tf t| d d
i qªW nê| jdkr¤x¬|D ]¤| _
dt| jj dt| j
j d}t tdddg }}xT|D ]L| _t d| d| j d¡ tf t| d d
i\}}}| || ¡ q"W tj||dd qÞW t ddd d g¡ t |d nt d | j d d S )!Nzrequirements.txt)Ú tensorboardÚthop)Úexclude)r rx r gü©ñÒMbP?u$ WARNING â ï¸ confidence threshold z! > 0.001 produces invalid resultsu` WARNING â ï¸ --save-hybrid will return high mAP from hybrid labels, not from predictions alonerf r )g Ð?gÍÌÌÌÌÌÜ?Fr( FZstudyZstudy_r© z.txté i é z
Running z --imgsz z...z%10.4g)ÚfmtrR z-rz study.zipz study_*.txt)rI z--task z2 not in ("train", "val", "test", "speed", "study")) r
rö r r r r× r$ rê r} rÙ r ræ r. rï Ú is_availabler] rÐ r r% r r rO rb r rS r^ ÚsavetxtÚ
subprocessr ÚNotImplementedError)r r r@ rI Úyr§ r© rÆ rA rA rB Úmain s2
" r Ú__main__)VÚ__doc__ry r rÝ r ÚsysÚpathlibr rg r^ r. r rû Úpandasr r Ú__file__ÚresolveÚFILEr~ rö rÛ rU rS ÚrelpathÚcwdÚ
models.commonr Úutils.callbacksr Úutils.dataloadersr Ú
utils.generalr r r
r r r
r r r r r r r r r r r r r Ú
utils.metricsr r r Ú utils.plotsr r r Úutils.torch_utilsr! r" Úutils.my_modelr# Útorch.nn.functionalÚnnÚ
functionalrñ ÚreÚtorchvision.opsr$ Zsklearn.metricsr% r& r' r( rC rZ rr rv rê r r Ú__name__r rA rA rA rB Ú<module> s
T
! S #