/j8ddlmZddlZddlZddlZddlZddlZddlmZddl m Z m Z ddl m Z ddlZddlZddlmZddlmZmZmZmZmZmZddlmZdd lmZmZdd lm Z d>d Z!d>dZ" d?d@dZ#dAd Z$dBd"Z%dCd&Z&dDd)Z'dEdFd/Z(d0Z)dGdHd6Z*dId9Z+dJdKd=Z,dS)L) annotationsN) defaultdict)ThreadPoolExecutor as_completed)Path)Image) ASSETS_URL DATASETS_DIRLOGGER NUM_THREADSTQDMYAML) check_file)download zip_directory)increment_pathreturn list[int]c gdS)aConvert 91-index COCO class IDs to 80-index COCO class IDs. Returns: (list[int | None]): A list of 91 elements where the index represents the 91-index class ID and the value is the corresponding 80-index class ID, or None if there is no mapping. )[r N NNN !"#$%&'N()*+,-./0123456789:;N<NN=N>?@ABCDEFGHNIJKLMNONre_/home/longshao/multi-rider-rag/.venv/lib/python3.11/site-packages/ultralytics/data/converter.pycoco91_to_coco80_classrhs\ \ \ \rfc gdS)aConvert 80-index (val2014) to 91-index (paper). Returns: (list[int]): A list of 80 class IDs where each value is the corresponding 91-index class ID. Examples: >>> import numpy as np >>> a = np.loadtxt("data/coco.names", dtype="str", delimiter="\n") >>> b = np.loadtxt("data/coco_paper.names", dtype="str", delimiter="\n") Convert the darknet to COCO format >>> x1 = [list(a[i] == b).index(True) + 1 for i in range(80)] Convert the COCO to darknet format >>> x2 = [list(b[i] == a).index(True) if any(b[i] == a) else None for i in range(91)] References: https://tech.amikelive.com/node-718/what-object-categories-labels-are-in-coco-dataset/ )Prrrrrrrrrrr r"r#r$r%r&r'r(r)r*r+r,r-r.r0r1r4r5r6r7r8r9r:r;r<r=r>r?r@rArCrDrErFrGrHrIrJrKrLrMrNrOrPrQrRrSrTrUrVrXr[r]r^r_r`rarbrcrdPQRTUVWXYZrererfrgcoco80_to_coco91_classrt~s(Q Q Q Qrf../coco/annotations/coco_converted/FT labels_dirstrsave_dir use_segmentsbool use_keypoints cls91to80lvisc  t|}|dz |dz fD]}|ddt}tt |dD]}|rdn|jdd} t |dz | z } | dd|r4| dz dd| d z ddt|d 5} tj | } d d d n #1swxYwYd | dD} tt}| dD]#}||d|$g}t|jd|D]R\}}| |d}|d|d}}|r5t#t |ddn|d} |r2|t#t d| z g}g}g}|D]}|ddrt)j|dt(j}|d dxx|dd dz z cc<|ddgxx|zcc<|dd gxx|zcc<|ddks |d dkr|r||d!dz n |d!dz }|g|}||vr|r|d"||t)j|d"d#d t)j||dgz d#z|||rW|d$}|t3|dkr|gt3|dkr{t5|}t)j|d%t)j||gz d#}||g|Td&|D}t)j|d#dt)j||gz d#}||g|t| | z d'd(d 5}t;t3|D]{}|r g||R}n.g|r!t3||dkr||n||R}|d)t3|z|zd*z| d d d n #1swxYwYT|rt ||j d+dd,d'z } t| d(d 5} | !d-|Dd d d n #1swxYwYtEj#|rd.nd/d0|d S)1aConvert COCO dataset annotations to a YOLO annotation format suitable for training YOLO models. Args: labels_dir (str, optional): Path to directory containing COCO dataset annotation files. save_dir (str, optional): Path to directory to save results to. use_segments (bool, optional): Whether to include segmentation masks in the output. use_keypoints (bool, optional): Whether to include keypoint annotations in the output. cls91to80 (bool, optional): Whether to map 91 COCO class IDs to the corresponding 80 COCO class IDs. lvis (bool, optional): Whether to convert data in lvis dataset way. Examples: >>> from ultralytics.data.converter import convert_coco Convert COCO annotations to YOLO format >>> convert_coco("coco/annotations/", use_segments=True, use_keypoints=False, cls91to80=False) Convert LVIS annotations to YOLO format >>> convert_coco("lvis/annotations/", use_segments=True, use_keypoints=False, cls91to80=False, lvis=True) labelsimagesTparentsexist_okz*.json instances_ train2017val2017utf-8encodingNc$i|] }|dd|S)iddre).0xs rg z convert_coco..s"<<<QtW..!<<z convert_coco..Gs% ; ; ;q ; ;A ; ; ; ;rf.txta%g  lvis_v1_z.jsonc3 K|] }|dV dSrNrerlines rg zconvert_coco..Ys(??T[[[??????rfLVISCOCOz/ data converted successfully. Results saved to )$rmkdirrhsortedrresolveglobstemreplaceopenjsonloadrlistappendr itemsrx relative_togetnparrayfloat64tolistreshapelenmerge_multi_segment concatenate with_suffixrangewriterstripname writelinesr info)!rwryrzr|r}r~pcoco80 json_filelnamefnfdatarrann image_txtimg_idannsimghwbboxessegmentsrboxclssegsfilerrfilenames! rg convert_cocors8h''H  (X"5 5-- t,,,,$ % %FD,,4466;;HEEFFO@O@ H  6 6|R H H (^^h & . ---  @+  $ $TD $ A A A )^ " "4$ " ? ? ? )g . . . !9Q<N>N!N W WXZ [ [ b b d dA$OOSI1I666rAv**622C'JJJ Kds6{{++KKA$1)A,11 -9ac(1+>N>NQR>R>RhqkkX^_`Xa  JJD 199;;dBTIJJJJK K K K K K K K K K K K K K K K  @H~~ (>(>z2(N(N(V(VW^`f(g(ggHhg666 @! ??Y?????? @ @ @ @ @ @ @ @ @ @ @ @ @ @ @ KT-66vss_g_o_o_q_qssttttts7D99D= D= BXX X Z  Z Z masks_dir output_dirclassesintc "dt|D}t|D]R}|jdvrEt jt |tj}|j\}}tj d|d|d|tj |}g} |D]1} | dkr | | d} | dkrtjd| d |d Bt j|| ktjtjtj\} } | D]}t)|d kr|}| g}|D]Z}|t/|d|z d |t/|d |z d [| |3t||jdz }t3|dd5}| D]B}dt7t |}||dzC dddn #1swxYwYtj d|d|d|TdS)uConvert a dataset of segmentation mask images to the YOLO segmentation format. This function takes the directory containing the binary format mask images and converts them into YOLO segmentation format. The converted masks are saved in the specified output directory. Args: masks_dir (str): The path to the directory where all mask images (png, jpg) are stored. output_dir (str): The path to the directory where the converted YOLO segmentation masks will be stored. classes (int): Total number of classes in the dataset, e.g., 80 for COCO. Examples: >>> from ultralytics.data.converter import convert_segment_masks_to_yolo_seg The classes here is the total classes in the dataset, for COCO dataset we have 80 classes >>> convert_segment_masks_to_yolo_seg("path/to/masks_directory", "path/to/output/directory", classes=80) Notes: The expected directory structure for the masks is: - masks ├─ mask_image_01.png or mask_image_01.jpg ├─ mask_image_02.png or mask_image_02.jpg ├─ mask_image_03.png or mask_image_03.jpg └─ mask_image_04.png or mask_image_04.jpg After execution, the labels will be organized in the following structure: - output_dir ├─ mask_yolo_01.txt ├─ mask_yolo_02.txt ├─ mask_yolo_03.txt └─ mask_yolo_04.txt ci|]}|dz| S)rrerrs rgrz5convert_segment_masks_to_yolo_seg..s???1a!eQ???rf>.jpg.png Processing z imgsz = z x rrzUnknown class for pixel value z in file z , skipping.rrrrrrr rNzProcessed and stored at )rriterdirsuffixcv2imreadrxIMREAD_GRAYSCALEshaper rruniquerwarning findContoursastypeuint8 RETR_EXTERNALCHAIN_APPROX_SIMPLErsqueezerroundrrjoinmapr)rrrpixel_to_class_mapping mask_pathmask img_height img_width unique_valuesyolo_format_datavalue class_indexcontours_contour yolo_formatpoint output_pathritemrs rg!convert_segment_masks_to_yolo_segr^s D@?g???)__,,..%e%e  / / /:c)nnc.BCCD$(J !J KTiTT*TTTT U U UIdOOM! & = =A::488CC "$$N#jE#j#jT]#j#j#jkkk".U]**2844c6GI` ! (==G7||q((")//"3"3'2m %,PPE'..uU1X 5I1/M/MNNN'..uU1X 5JA/N/NOOOO(// <<<=z** -D-D-DDKk3999 ,T,,,D88CTNN33DJJtd{++++, , , , , , , , , , , , , , , , Kc;ccccXacc d d dK%e%esAI%%I) ,I) dota_root_pathc . t|}iddddddddd d d d d ddddddddddddddddddd d!d"d#d$i d9 fd- }d.D]}|d/z |z }|d0z |d1z }|d0z |z }|d2d23t|}t |d4|d56D]W}|jd7kr|j}tjt|} | j d8d\} } ||| | ||Xd8S):u Convert DOTA dataset annotations to YOLO OBB (Oriented Bounding Box) format. The function processes images in the 'train' and 'val' folders of the DOTA dataset. For each image, it reads the associated label from the original labels directory and writes new labels in YOLO OBB format to a new directory. Args: dota_root_path (str): The root directory path of the DOTA dataset. Examples: >>> from ultralytics.data.converter import convert_dota_to_yolo_obb >>> convert_dota_to_yolo_obb("path/to/DOTA") Notes: The directory structure assumed for the DOTA dataset: - DOTA ├─ images │ ├─ train │ └─ val └─ labels ├─ train_original └─ val_original After execution, the function will organize the labels into: - DOTA └─ labels ├─ train └─ val planershiprz storage-tankrzbaseball-diamondrz tennis-courtrzbasketball-courtrzground-track-fieldrharborrbridgerz large-vehiclerz small-vehicler helicopterr roundaboutr!zsoccer-ball-fieldr"z swimming-poolr#zcontainer-craner$airportr%helipadr& image_namerx image_widthr image_heightorig_label_dirrryc ||dz }||dz }|d5}|d5}|} | D]} | } t | dkr<| d} | } d| ddDfdt dD}d |D}|| d d |d  dddn #1swxYwYddddS#1swxYwYdS) zaConvert a single image's DOTA annotation to YOLO OBB format and save it to a specified directory.rrrrrc,g|]}t|Sre)float)rrs rgrzCconvert_dota_to_yolo_obb..convert_label..s666q%((666rfNcNg|]!}|dzdkr |z n |z "S)rrre)rrcoordsrrs rgrzCconvert_dota_to_yolo_obb..convert_label..sI%%%\]q1uzzF1I ++vay.convert_label..s#R#R#RuuNN#R#R#Rrfrr)r readlinesstripsplitrrrr)rrrr ryorig_label_path save_pathrglinesrparts class_name class_idxnormalized_coordsformatted_coordsr& class_mappings `` @rg convert_labelz/convert_dota_to_yolo_obb..convert_labels+(j+>+>+>>*2222  ! !# & & G!Y^^C-@-@ GAKKMME G G **,,u::>>"1X )*5 66E"1"I666%%%%%%afghaiai%%%!$S#R@Q#R#R#R 9EEsxx0@'A'AEEEFFFF G G G G G G G G G G G G G G G G G G G G G G G G G G G G G G G G G Gs6D4CD D4D D4#D $D44D8;D8>valtrainrr _originalTrrz imagesrrN) rrxrrrrr rryr) rrrrr rrrrrxr) rr6phase image_dirr ry image_paths image_pathimage_name_without_extrrrr5s @rgconvert_dota_to_yolo_obbr?s'>.))N  A    A  a ! !   b b R  2!" 2#$ 2%M*GGGGGG("RR"X-5 '(25H5H5HH!H,u4td3339,,..// {1Mu1M1M1MNNN R RJ F**%/_ "*S__--C9RaR=DAq M0!Q Q Q Q Q  RRRrfarr1 np.ndarrayarr2c|dddddf|dddddfz dzd}tjtj|d|jS)aFind a pair of indexes with the shortest distance between two arrays of 2D points. Args: arr1 (np.ndarray): A NumPy array of shape (N, 2) representing N 2D points. arr2 (np.ndarray): A NumPy array of shape (M, 2) representing M 2D points. Returns: (tuple[int, int]): A tuple (idx1, idx2) where idx1 is the index in arr1 and idx2 is the index in arr2 of the pair with the shortest distance. Nrrr)sumr unravel_indexargminr)r@rBdiss rg min_indexrHsk D!!! tD!!!QQQJ/ /A 5 : :2 > >C BIc555sy A AArfr list[list]cg}d|D}dtt|D}tdt|D]]}t||dz ||\}}||dz ||||^tdD]}|dkr,t |D]\}}t|dkr5|d|dkr#|ddd}||dddddf||<t j|||d d||<t j||||ddg||<|dt|dz hvr|||d|d|dz g}||||d|ddz6tt|dz ddD]a}|dt|dz hvrI||}t|d|dz }||||db|S) aMerge multiple segments into one list by connecting the coordinates with the minimum distance between each segment. This function connects these coordinates with a thin line to merge all segments into one. Args: segments (list[list]): Original segmentations in COCO's JSON file. Each element is a list of coordinates, like [segmentation1, segmentation2,...]. Returns: (list[np.ndarray]): A list of connected segments represented as NumPy arrays. c^g|]*}tj|dd+S)rr)rrrrs rgrz'merge_multi_segment.."s0===q ##B**===rfcg|]}gSrere)rr s rgrz'merge_multi_segment..#s111q111rfrrrNrr) rrrHr enumeraterrollrabs) rridx_listridx1idx2kidxnidxs rgrrs A==H===H11E#h--00111H1c(mm $ $!!xA << dQt$$$ 4    1XX11 66#H-- ? ?3s88q==SVc!f__ddd)C"*1+dddAAAg"6HQK ghqkCF7CCC  nhqk8A;rr?-KLL CMMA-...HHXa[))))c!fs1vo.CHHXa[Q#a&1*)<=>>>> ? 3x==1,b"55 1 1QH  1222"1+Cs1vA//DHHXa[/000  1 Hrfsam_b.ptim_dir str | Pathstr | Path | None sam_modelc ddlm}ddlm}ddlm}||t ttdd}t|j dd dkrtj d d Stj d ||}t|j t|j d D]}|d\} } |d} t| dkr)| d d ddgfxx| zcc<| d d ddgfxx| zcc<tj|d} || || dd|} | djj|d <|rt%|nt%|jdz }|dd|j D]}g}t%|ddj}||z }|d}t/|d D]\}}t|dkrt1||g|dR}|dt|z|zt9|dd5}|d |Dd d d n #1swxYwYtj d!|d S)"uConvert existing object detection dataset (bounding boxes) to segmentation dataset in YOLO format. Generates segmentation data using SAM auto-annotator as needed. Args: im_dir (str | Path): Path to image directory to convert. save_dir (str | Path, optional): Path to save the generated labels, labels will be saved into `labels-segment` in the same directory level of `im_dir` if save_dir is None. sam_model (str): Segmentation model to use for intermediate segmentation data. device (int | str, optional): The specific device to run SAM models. Notes: The input directory structure assumed for dataset: - im_dir ├─ 001.jpg ├─ ... └─ NNN.jpg - labels ├─ 001.txt ├─ ... └─ NNN.txt r)SAM) YOLODataset) xywh2xyxyir)nameschannels)rrz;Segmentation labels detected, no need to generate new ones!NzBDetection labels detected, generating segment labels by SAM model!zGenerating segment labelstotalrrrrrim_fileF)rverbosesavedevicezlabels-segmentTrrrrrrrrc3 K|] }|dzV dSrre)rtexts rgrz$yolo_bbox2segment..s&77777777rfz"Generated segment labels saved in ) ultralyticsr\ultralytics.datar]ultralytics.utils.opsr^dictrrrrr rr rrmasksxynrparentrrrrMrrrrrr)rWryrZrfr\r]r^datasetlabelrrboxesim sam_resultstextslb_nametxt_filerrrrrs rgyolo_bbox2segmentrxGsf0 ,,,,,,//////k&t$uT{{2C2Ca'P'P'PQQQG 7>! Z ())A-- QRRR KTUUUIIgnC,?,?Fabbb 5 5W~1h u::??  aaa!QiA aaa!QiA Zi( ) )i99U+;+;UQV_efff 'N04j!)UtH~~~tF||/BEU/UH NN4$N/// 8 8uY'((44V<<Ag%EleJ/00 > >DAq1vv{{AKK0!))B--00D LL%#d))+3355< = = = = (C' 2 2 2 8a LL77777 7 7 7 8 8 8 8 8 8 8 8 8 8 8 8 8 8 8 K?X??@@@@@s KK K c  ddtdz ttdgjt jdz dz d t t 5 d D]}d z |z }|d d |dz }| ryt|d5}fd|D}dddn #1swxYwY fd|D}tt|t|d|D]}tjd|d|d dddn #1swxYwYtjddS)aCreate a synthetic COCO dataset with random images based on filenames from label lists. This function downloads COCO labels, reads image filenames from label list files, creates synthetic images for train2017 and val2017 subsets, and organizes them in the COCO dataset structure. It uses multithreading to generate images efficiently. Examples: >>> from ultralytics.data.converter import create_synthetic_coco_dataset >>> create_synthetic_coco_dataset() Notes: - Requires internet connection to download label files. - Generates random RGB images of varying sizes (480x480 to 640x640 pixels). - Existing test2017 directory is removed as it's not needed. - Reads image filenames from train2017.txt and val2017.txt files. image_filerc R|stjddtjddf}tjd|tjddtjddtjddf|dSdS)zcGenerate a synthetic image with random size and color for dataset augmentation or testing purposes.iiRGBr)sizecolorN)existsrandomrandintrnewre)rzr~s rgcreate_synthetic_imagez=create_synthetic_coco_dataset..create_synthetic_images  "" N3,,fnS#.F.FGD I~a--v~a/E/Ev~VWY\G]G]^   d:   rfcocoz/coco2017labels-segments.zip)dirrtest2017T ignore_errors) max_workers>rrrrrrrc>g|]}|z Sre)r*)rrrs rgrz1create_synthetic_coco_dataset..s&"D"D"D$3#5"D"D"DrfNc<g|]}|Sre)submit)rrzrexecutors rgrz1create_synthetic_coco_dataset..s)mmmS]8??+A:NNmmmrfzGenerating images for raz Labels file z- does not exist. Skipping image creation for .z,Synthetic COCO dataset created successfully.)rzr)r rr roshutilrmtreerr rrrr rrr rr) subset subset_dirlabel_list_filer image_filesfuturesr rrrs @@@rgcreate_synthetic_coco_datasetrsm$  C 999 : KKKK M#.:-TBBBB  4 4 4w. w wFx&0J   TD  9 9 9"vOOO3O%%'' w/G<<<E"D"D"D"D!"D"D"DKEEEEEEEEEEEEEEEnmmmmalmmml7333w<?????s8"AE3C EC EC A!EE Erpath n_channelsrzipcddlm}ddlm}t rfd|ddhz D}|D]Z} t |||r|*#t$r$}tj d|d|Yd }~Sd }~wwxYw|rtd Sd S d } tjtjt!tj} t%jgd } t%jd d |} || j| ddd} | | }tjt!| t%j|ddt$jdddtj d| d S)aKConvert RGB images to multispectral images by interpolating across wavelength bands. This function takes RGB images and interpolates them to create multispectral images with a specified number of channels. It can process either a single image or a directory of images. Args: path (str | Path): Path to an image file or directory containing images to convert. n_channels (int): Number of spectral channels to generate in the output image. replace (bool): Whether to replace the original image file with the converted one. zip (bool): Whether to zip the converted images into a zip file. Examples: Convert a single image >>> convert_to_multispectral("path/to/image.jpg", n_channels=10) Convert a dataset >>> convert_to_multispectral("coco8", n_channels=10) r)interp1d) IMG_FORMATScJg|]}d|D]}| S)z*.)rglob)rextrrs rgrz,convert_to_multispectral..s<aaa#$**U_Z]U_U_J`J`aaQAaaaarftiftiffzError converting z: Nz.tiff)iiiiilinearF extrapolate)kind bounds_error fill_valuer}rrz Converted )scipy.interpolaterultralytics.data.utilsrris_dirconvert_to_multispectralunlink Exceptionr rrrrcvtColorrrx COLOR_BGR2RGBrrlinspaceT imwritemulticliprr transpose)rrrrrrim_filesim_patherrrgb_wavelengthstarget_wavelengthsr multispectrals` rgrrs &+*****222222 ::D {{}}0aaaa+"?aaa @ @G @(*===%NN$$$ @ @ @ >>>1>>???????? @  $       &&w// l3:c$ii00#2CDD(???33[c:>> H_&([h i i i,--  [))27=!S+I+I+P+PQSQY+Z+Z+d+defhikl+m+mnnn .../////s &A00 B:BB image_recordsrcjg}g}|D]}|didgD]a}t|dz }|dkr2||||ddt|dkrnbt|dkrn|r tt|dkrt d|d}|d zdkr t d |Dr|d zd gS|d zdkr|d zdkr|d zd gSt d) aInfer kpt_shape [num_keypoints, dims] from NDJSON pose annotations. Scans up to 50 pose annotations across image records. Annotation format is [classId, cx, cy, w, h, kp1_x, kp1_y, kp1_vis, ...] so keypoint values start at index 5. Tries dims=3 first (x, y, visibility) with visibility validation ({0, 1, 2}), then falls back to dims=2 (x, y only) when values are unambiguously not divisible by 3. rposerrNrGrzZPose dataset missing required 'kpt_shape'. See https://docs.ultralytics.com/datasets/pose/rc3:K|]}|dddD]}|dvV dS)rNr)rrrre)rrvs rgrz*_infer_ndjson_kpt_shape.. s=KKQ1QTT7KKa!y.KKKKKKKrfr)rrrset ValueErrorall)r kpt_lengthssamplesrecordrkpt_lenns rg_infer_ndjson_kpt_shaperswKG  ::mR0044VR@@  C#hhlG{{""7+++s122w''';2%%& {  r ! ! E " w#c+..//144uvvvAA 1uzzcKK'KKKKKzQ{ 1uzza!eqjjQ{ q r rrrf ndjson_pathrrc `  !"#$Kddlm}|dddltt |}t|pt }t |5}d|D}dddn #1swxYwY|d|dd}}tj}|D]Y}| tj d| Dd  Z|dd } ||jz d z } | r? t#j| d | kr| Sn#t($rYnwxYwd |D} |ddkd|di Dt+|dd} sd| vrt-dt/| d| vrd| vrd|D} t+| dkr t-dt+| dt1jd| t7dt+| dz}| d|D]}d|d<| dt;jdt+| dt+| |z d|d | d!krd"|vrt?||d"< r1| !d#stEj#d$z d%$dd&d}stK|}|d'<|&dd|&d(dt/| D]D}d)z |z $dd&d$z |z $dd&d*|||<Efd+!tOj(tSd,t+|"*4d{V#tWt+|d-|j,d.d/t+|d01 !"#fd2$tOj-$fd3|Dd{V} .dddd{Vn#1d{VswxYwYt_d4|D}|dkrtad5|d6|t+|kr*t;jd7|d8t+|d9|rtc}|D]}|d|d:}}rx|d;i}|daConvert NDJSON dataset format to Ultralytics YOLO dataset structure. This function converts datasets stored in NDJSON (Newline Delimited JSON) format to the standard YOLO format. For detection/segmentation/pose/obb tasks, it creates separate directories for images and labels. For classification tasks, it creates the ImageNet-style {split}/{class_name}/ folder structure. It supports parallel processing for efficient conversion of large datasets and can download images from URLs. The NDJSON format consists of: - First line: Dataset metadata with class names, task type, and configuration - Subsequent lines: Individual image records with annotations and optional URLs Args: ndjson_path (str | Path): Path to the input NDJSON file containing dataset information. output_path (str | Path | None, optional): Directory where the converted YOLO dataset will be saved. If None, uses the DATASETS_DIR directory. Defaults to None. Returns: (Path): Path to the generated data.yaml file (detection) or dataset directory (classification). Examples: Convert a local NDJSON file: >>> yaml_path = await convert_ndjson_to_yolo("dataset.ndjson") >>> print(f"Dataset converted to: {yaml_path}") Convert with custom output directory: >>> yaml_path = await convert_ndjson_to_yolo("dataset.ndjson", output_path="./converted_datasets") Use with YOLO training >>> from ultralytics import YOLO >>> model = YOLO("yolo26n.pt") >>> model.train(data="https://github.com/ultralytics/assets/releases/download/v0.0.0/coco8-ndjson.ndjson") r)check_requirementsaiohttpNcg|]<}|tj|=Sre)r*rloadsrs rgrz*convert_ndjson_to_yolo..@s7HHHd4::<<HDJJLL))HHHrfrc&i|]\}}|dk ||S)urlrerrSrs rgrz*convert_ndjson_to_yolo..Fs#GGGtq!AJJaJJJrfT) sort_keysr%z data.yamlhashch|] }|d S)r+re)rrs rg z)convert_ndjson_to_yolo..Rs : : :&fWo : : :rftaskclassifyc4i|]\}}t||Sre)rrs rgrz*convert_ndjson_to_yolo..Vs$WWWA3q661WWWrf class_namesdetectr8z6Dataset missing required 'train' split. Found splits: r7testcDg|]}|ddk|S)r+r8)rrr"s rgrz*convert_ndjson_to_yolo.._s,SSS1w79R9RQ9R9R9RrfrzDataset has only zR image(s) and no 'val' split. Need at least 2 images to auto-split into train/val.rr+u?WARNING ⚠️ No 'val' split found in dataset. Auto-splitting z images into z train, zS val. For best results, manually assign validation images in Platform dataset page.r kpt_shape) missing_okrrrr_typerzimages/c <K|4d{V|d|d}}|di}rL|jdg}|r|dnd}|t|}|z |z |z } ndz |z |z } dz |z t|jd z } g} |jD]} d || D} | | rd | d znd | srkd D]h} | |kr r | z |z |z n dz | z |z }|r3| j dd| | ni| s_|dx}rG| j ddtdD]} || d4d{V}| | |d{Vdddd{Vn#1d{VswxYwYdddd{VdS#t $ra}|dkrt#jd|zd{Vn2t'jd|d|Yd}~dddd{VdSYd}~d}~wwxYw dddd{VdS#1d{VswxYwYdS)z/Process single image record with async session.Nr+rrclassificationrrrrc^g|]*}dtt|+S)r)rrrx)rrs rgrzBconvert_ndjson_to_yolo..process_record..s,%\%\%\4chhs3~~&>&>%\%\%\rfrr)r8r7rTrrrr3)rb)timeoutrzFailed to download z after 3 attempts: F)rrxrrkeys write_textrrrorrenamer ClientTimeoutraise_for_status write_bytesreadrasynciosleepr r)session semaphorerr+ original_namer class_idsclass_idr1r= label_pathlines_to_writekeyr candidatehttp_urlattemptresponser_reuserr dataset_diris_classifications rgprocess_recordz.convert_ndjson_to_yolo..process_records|2 2 2 2 2 2 2 2 #)'?F6N=E **]B77K  b+KO,_>_>__ !#+;+--C%\%\;WZK[%\%\%\N%%.&`dii&?&?$&F&F^`aaa$$&& - "5 " "::$ 1N[1_z9MII"-"81"<}"L" %++--"&-33D43PPP%,,Z888!E""((**-FJJuK/L )L /K44L  LLz Converting u → z (z images)rac\K|d{V}d|S)Nr)update)rresultpbarrrrs rgtracked_processz/convert_ndjson_to_yolo..tracked_processs@)>'9fEEEEEEEEF KKNNNMrfc&g|] }|Srere)rrrs rgrz*convert_ndjson_to_yolo..s#(](](]V)@)@(](](]rfc3K|]}|dV dS)rNrers rgrz)convert_ndjson_to_yolo..s'00aa0000000rfz#Failed to download any images from z$. Check network connection and URLs.z Downloaded /z images from rrr*)5ultralytics.utils.checksrrrrr rhashlibsha256rrdumpsrencode hexdigestris_filerrrrrrrrRandomshufflemaxaddr rrrrrrrrlpopr Semaphoremin ClientSessionr rgathercloserD RuntimeErrorrrxrre)%rrrrr/dataset_recordr_hr"_hash yaml_pathsplitsr train_records val_count data_yamlr+results success_countexpected_pathsrrrcidscidimg_rootrrrrrrrrrrrs% @@@@@@@@@@rgconvert_ndjson_to_yolor(sB<;;;;;y!!!NNNz+..//K{2l33K k  IaHHaHHHIIIIIIIIIIIIIII$)!HeABBiMN   B cc $*GGqwwyyGGGSWXXX__aabbbb LLNN3B3 E 00Kk)I y##''//588  9    D  : :M : : :F'**622j@WW););M2)N)N)T)T)V)VWWWK    fh / /D  & fV\]cVdVdffgg g   6#7#7SS SSSM=!!A%% LM(:(:LLL M!   $ $] 3 3 3As=11R788I":I:. # #"' JJu    Na"%m"4"4aaCF}CUCUXaCaaaktaaa     v~~+^;;&=m&L&L{#   ! !F FD)))  F M+0 E E E EdT222I  1(( ( ' mT*** fd###F^^ 1 1E 8 #e + 2 24$ 2 O O O 8 #e + 2 24$ 2 O O O000Ie  444444444n!#c3}+=+=">">??I$$&&       'm$$a{/aakaaSEWEWaaa              (](](](]}(](](]^^^^^^^                            00700000Mrrrrssss=))))c]ccS5G5GccVaccddd F FAj!F)tA  FeeM2..ww/44!%,d1gg1"";?[__S#c((5S5S#SVZ#Z[[[["";#9A#=#DEEEE"3Q;;+:P$$  Ayy{{ q66 " & )Y'''s7# A<<BB,E66 FFr2su#""""" ######???????? WWWWWWWWWWWWWWWW//////????????222222ccccLeeeeR-% uuuuuuuuuupHeHeHeHeVYRYRYRYRx B B B B0 0 0 0 f=A=A=A=A=A@4@4@4@n0000000000f#s#s#s#sLTTTTTTTrf