o )i](@sddlZddlZddlZddlZddlmZddlmZddlmZddl m Z m Z m Z m Z mZddlmZddlmZmZdd lmZmZmZmZdd lmZdd lmZd ed ededdfddZGdddeZdS)N)partial)Pool)path)AnyCallableDictOptionalTuple)Tensor) find_classes make_dataset)check_integritydownload_and_extract_archive download_urlverify_str_arg) VideoClips) VisionDatasettarpath videopathlinereturncCst|||dSN)r)rrrrl/var/www/html/eduruby.in/lip-sync/lip-sync-env/lib/python3.10/site-packages/torchvision/datasets/kinetics.py_dl_wrapsrc)s,eZdZdZddddZddddZ        d5dededededeededee de edfde dededee ee fdeded ed!ed"ed#e d$ed%d f(fd&d' Zd6d(d)Zd6d*d+Zd6d,d-Zed%e ee ffd.d/Zd%efd0d1Zd2ed%e eeeffd3d4ZZS)7Kineticsuz `Generic Kinetics `_ dataset. Kinetics-400/600/700 are action recognition video datasets. This dataset consider every video as a collection of video clips of fixed size, specified by ``frames_per_clip``, where the step in frames between each clip is given by ``step_between_clips``. To give an example, for 2 videos with 10 and 15 frames respectively, if ``frames_per_clip=5`` and ``step_between_clips=5``, the dataset size will be (2 + 3) = 5, where the first two elements will come from video 1, and the next three elements from video 2. Note that we drop clips which do not have exactly ``frames_per_clip`` elements, so not all frames in a video might be present. Args: root (string): Root directory of the Kinetics Dataset. Directory should be structured as follows: .. code:: root/ ├── split │ ├── class1 │ │ ├── clip1.mp4 │ │ ├── clip2.mp4 │ │ ├── clip3.mp4 │ │ ├── ... │ ├── class2 │ │ ├── clipx.mp4 │ │ └── ... Note: split is appended automatically using the split argument. frames_per_clip (int): number of frames in a clip num_classes (int): select between Kinetics-400 (default), Kinetics-600, and Kinetics-700 split (str): split of the dataset to consider; supports ``"train"`` (default) ``"val"`` ``"test"`` frame_rate (float): If omitted, interpolate different frame rate for each clip. step_between_clips (int): number of frames between each clip transform (callable, optional): A function/transform that takes in a TxHxWxC video and returns a transformed version. download (bool): Download the official version of the dataset to root folder. num_workers (int): Use multiple workers for VideoClips creation num_download_workers (int): Use multiprocessing in order to speed up download. output_format (str, optional): The format of the output video tensors (before transforms). Can be either "THWC" or "TCHW" (default). Note that in most other utils and datasets, the default is actually "THWC". Returns: tuple: A 3-tuple with the following entries: - video (Tensor[T, C, H, W] or Tensor[T, H, W, C]): the `T` video frames in torch.uint8 tensor - audio(Tensor[K, L]): the audio frames, where `K` is the number of channels and `L` is the number of points in torch.float tensor - label (int): class of the video clip Raises: RuntimeError: If ``download is True`` and the video archives are already extracted. zChttps://s3.amazonaws.com/kinetics/400/{split}/k400_{split}_path.txtzChttps://s3.amazonaws.com/kinetics/600/{split}/k600_{split}_path.txtzMhttps://s3.amazonaws.com/kinetics/700_2020/{split}/k700_2020_{split}_path.txt400Z600Z700z=https://s3.amazonaws.com/kinetics/400/annotations/{split}.csvz=https://s3.amazonaws.com/kinetics/600/annotations/{split}.csvzBhttps://s3.amazonaws.com/kinetics/700_2020/annotations/{split}.csvrtrainNr ZaviZmp4FrTCHWrootframes_per_clip num_classessplit frame_ratestep_between_clips transform extensions.downloadnum_download_workers num_workers_precomputed_metadata _video_width _video_height_video_min_dimension_audio_samples_audio_channels_legacy output_formatrcst|dgdd|_||_| |_||_||_|r+td||_d|_d}| r*t dnt |||_t|dgd d|_| rB| t |jt|j\|_}t|j||dd |_d d |jD}t||||| | | |||||d |_||_dS)Nr$r)argZ valid_valueszUsing legacy structureunknownZTHWCz2Cannot download the videos using legacy_structure.r%)rvaltest)Z is_valid_filecSsg|]}|dqS)rr).0xrrr sz%Kinetics.__init__..)r,r.r/r0r1r2r4)rr$r)r+r"r3print split_folderr% ValueErrorrjoindownload_and_process_videossuper__init__r classesr samplesr video_clipsr()selfr"r#r$r%r&r'r(r)r*r+r,r-r.r/r0r1r2r3r4Z class_to_idxZ video_list __class__rrrB[sH zKinetics.__init__cCsbt}|t}td||d|t}td||dtd||ddS)zEDownloads all the videos to the _root_ folder in the expected format.z%Elapsed time for downloading in mins <z$Elapsed time for processing in mins zElapsed time overall in mins N)time_download_videosr<_make_ds_structure)rFZticZtocZtoc2rrrr@sz$Kinetics.download_and_process_videosc Cst|jrtd|jdt|jd}t|jd}|j|jj|j d}t|t |}t |s:t ||t |}dd|D}Wdn1sTwY|jd krl|D] }t|||jq`dStt||j}t|j} | ||dS) adownload tarballs containing the video to "tars" folder and extract them into the _split_ folder where split is one of the official dataset splits. Raises: RuntimeError: if download folder exists, break to prevent downloading entire dataset again. zThe directory z[ already exists. If you want to re-download or re-extract the images, delete the directory.Ztarsfilesr%cSsg|] }tjj|ddqS)z/,:)safe)urllibparsequote)r9rrrrr;sz-Kinetics._download_videos..Nr )rexistsr= RuntimeErrorr?r" _TAR_URLSr$formatr%basenamerropenread splitlinesr+rrrrmap) rFZtar_pathZfile_list_pathZ split_urlZsplit_url_filepathfileZlist_video_urlsrpartZpoolprocrrrrKs(      zKinetics._download_videosc Cs0t|jd}tt||jds!t|j|jj|jd|t||jd}d}t |]}t |}|D]L}|j|dt |dt |dd}|d  d d  d d  dd  dd }tjt|j|ddt|j|} t| rt | t|j||q9WddS1swYdS)umove videos from split_folder/ ├── clip1.avi ├── clip2.avi to the correct format as described below: split_folder/ ├── class1 │ ├── clip1.avi annotationsz.csvrNz{ytid}_{start:06}_{end:06}.mp4Z youtube_idZ time_startZtime_end)Zytidstartendlabel _'()T)exist_okN)rr?r"rr%r_ANNOTATION_URLSr$rVrXcsv DictReaderintreplaceosmakedirsr=isfile) rFZannotation_pathr^Z file_fmtstrZcsvfilereaderrowfraZdownloaded_filerrrrLs0     ( "zKinetics._make_ds_structurecCs|jjSr)rEmetadatarFrrrrtszKinetics.metadatacCs |jSr)rEZ num_clipsrurrr__len__s zKinetics.__len__idxcCs@|j|\}}}}|j|d}|jdur||}|||fS)Nr )rEZget_cliprDr()rFrwZvideoZaudioinfoZ video_idxrarrr __getitem__s    zKinetics.__getitem__)rrNr Nr Fr r NrrrrrFr!)rN)__name__ __module__ __qualname____doc__rUristrrlrrr boolrrrBr@rKrLpropertyrtrvr ry __classcell__rrrGrrs:       B #$r) rjrnrJrP functoolsrmultiprocessingrrtypingrrrrr Ztorchr folderr r utilsrrrrZ video_utilsrZvisionrr~rrrrrrs