j;HddlZddlZddlZddlZddlZddlmZddlmZm Z m Z m Z m Z m Z ddlZddlZddlmZddlmZ ddlZejejjeejjjds ed Z ejZn#e$r ej ZYnwxYwn#e$red ZYnwxYwd3d Z!d e"fd Z#da$da% d4de&dej'de(de&de e e&efde ej'de e(de e&de e e&efd dfdZ)ddde(de(de&dd d!e e&e e e*e e*d"fe e*ffd e d#fd$Z+d%ej'd&e d#d'e*d(e(d ej'f d)Z, d5de&d,e e(efd-e e e(efde&d.e&d e ej'ej'e e&efff d/Z-ddd e"fd0Z.ddd e e*fd1Z/d6de&de&d e e e*e e(ffd2Z0dS)7N)Fraction)AnyDictListOptionalTupleUnion)_log_api_usage_once) _video_opt pict_typea+Your version of PyAV is too old for the necessary video operations in torchvision. If you are on Python 3.5, you will have to build from source (the conda-forge packages are not up-to-date). See https://github.com/mikeboers/PyAV#installation for instructions on how to install PyAV on your system. zPyAV is not installed, and is necessary for the video operations in torchvision. See https://github.com/mikeboers/PyAV#installation for instructions on how to install PyAV on your system. returncHtttrtdSN isinstanceav ExceptionEC:\Projects\Historical\venv\Lib\site-packages\torchvision/io/video.py_check_av_availabler+s#"i  rc8ttt Srrrrr _av_availabler0s"i(( ((r libx264filename video_arrayfps video_codecoptions audio_array audio_fps audio_codec audio_optionsc Jtjs2tjst t t tj|tj d}t|trtj |}tj|d5} | ||} |jd| _|jd| _|d krd nd | _|pi| _|7d d ddddddddd } | ||} |pi| _|jd} | dkrdnd}| jjdjj}tj| |}tj| d|}tj|||}||_| |D]}| || D]}| ||D]{}tj!|d } d|_"n"#tF$rddl$m%}|j&|_"YnwxYw| |D]}| ||| D]}| | d d d d S#1swxYwYd S)aM Writes a 4d tensor in [T, H, W, C] format in a video file. This function relies on PyAV (therefore, ultimately FFmpeg) to encode videos, you can get more fine-grained control by referring to the other options at your disposal within `the FFMpeg wiki `_. .. warning:: In the near future, we intend to centralize PyTorch's video decoding capabilities within the `torchcodec `_ project. We encourage you to try it out and share your feedback, as the torchvision video decoders will eventually be deprecated. Args: filename (str): path where the video will be saved video_array (Tensor[T, H, W, C]): tensor containing the individual frames, as a uint8 tensor in [T, H, W, C] format fps (Number): video frames per second video_codec (str): the name of the video codec, i.e. "libx264", "h264", etc. options (Dict): dictionary containing options to be passed into the PyAV video stream. The list of options is codec-dependent and can all be found from `the FFMpeg wiki `_. audio_array (Tensor[C, N]): tensor containing the audio, where C is the number of channels and N is the number of samples audio_fps (Number): audio sample rate, typically 44100 or 48000 audio_codec (str): the name of the audio codec, i.e. "mp3", "aac", etc. audio_options (Dict): dictionary containing options to be passed into the PyAV audio stream. The list of options is codec-dependent and can all be found from `the FFMpeg wiki `_. Examples:: >>> # Creating libx264 video with CRF 17, for visually lossless footage: >>> >>> from torchvision.io import write_video >>> # 1000 frames of 100x100, 3-channel image. >>> vid = torch.randn(1000, 100, 100, 3, dtype = torch.uint8) >>> write_video("video.mp4", options = {"crf": "17"}) dtypeT)forcew)mode)rater r libx264rgbyuv420prgb24NzNONE) PictureType)'torchjit is_scripting is_tracingr write_videor as_tensoruint8numpyrfloatnproundropen add_streamshapewidthheightpix_fmtr"streamsaudior>namer)astype AudioFrame from_ndarray sample_rateencodemux VideoFramer TypeErrorav.video.framerAr@)rrr r!r"r#r$r%r& containerstreamaudio_format_dtypesa_stream num_channels audio_layoutaudio_sample_fmt format_dtypeframepacketimgrAs rrFrF9sj 9 ! ! # #)EI,@,@,B,B)K(((/+U[AAAGGdGSSK#uhsmm  $ $ $6" %%k%<<"(+ #)!, &1\&A&Aw B  " # # !++Ki+HHH,2H &,Q/L'3a'7'788VL(06q9@E 8$78H$IJJL/+66<<4<HHOOP\]]KM..{CS\h.iiE )E "//%00 & & f%%%%"//++ & & f%%%% & &CM..s7.CCE 3"( 3 3 3666666"-"2 3 !--.. & & f%%%% &mmoo " "F MM& ! ! ! ! "k6"6"6"6"6"6"6"6"6"6"6"6"6"6"6"6"6"6"s8F3LJ LJ-*L,J--ALLLr_zav.container.Container start_offset end_offsetpts_unitr`zav.stream.Stream stream_name.zav.frame.Framectdz attztdz krtj|dkrlt t jd|jz ztdkr,t t j d|jz zntj did}d}|j dkrx|j j}|rjd|vrf|d} || d} t!jd | } | t!jd | } | | d d k}} t'| dz d } |rt'| |z d } || dd|n#t*$rgcYSwxYwd } t-|jdi|D]*\}}||j<|jkr|r | |kr| dz } (n+n#t*$rYnwxYwfdt3D}t5d krVd krPvrLfdD}t5|d kr+t'|}|d ||S)Nr secinfzBThe pts_unit 'pts' gives wrong results. Please use pts_unit 'sec'.TvideosDivXsDivX(\d+)Build(\d+)(\w)sDivX(\d+)b(\d+)(\w)prF) any_framebackwardr`cVg|]%}|jcxkrknn|&Srpts).0irkframesrjs r z%_read_from_stream..sE _ _ _A<6!9=3^3^3^3^T^3^3^3^3^3^fQi3^3^3^rc g|] }|k| Srr)rzr{rjs rr}z%_read_from_stream..s#BBB!\1A1AA1A1A1Arr) _CALLED_TIMES_GC_COLLECTION_INTERVALgccollectintmathfloor time_baserJceilwarningswarntype codec_context extradatafindresearchgroupmaxseek FFmpegError enumeratedecoderysortedleninsert)r_rjrkrlr`rm should_buffermax_buffer_sizerposdo seek_offset buffer_count_idxrgresultpreceding_framesfirst_frame_ptsr|s `` @r_read_from_streamrsQM..2IA2MMM 54:la&:J6J&KLLMM u % %TYzQ9I5I'JKKLLJ Z[[[ FMO {g(2   3I--..))C#$$A 5q99AyI5q99} ! d 2 KkAoq))K<+7;; {ed6RRRR  L  $%5Y%5%D%D %D%DEE  KD% %F59 yJ&& \O%C%C A%L '      ` _ _ _ _ _ _ _ _F 6{{Q ""rryTHWC start_ptsend_pts output_formatc tjs2tjst t |}|dvrtd|dddlm }|dkrNtj |std|tj||||\}}}n~t!|t#d }||krtd |d |i}g} g} tj} t'j|d 5} | jjr| jjdj} | jjrQt3| |||| jjdddi} | jjdj} | t#| |d<| jjr@t3| |||| jjdddi} | jjdj|d<dddn #1swxYwYn#t8$rYnwxYwd| D}d| D}|r'tjt=j|}n tj dtj!}|rt=j"|d}tj|}|dkratGtIj%|d| z z}|t#d kr'tGtIj&|d| z z}tO|| ||}n tj dtj(}|dkr|)dddd}|||fS)a Reads a video from a file, returning both the video frames and the audio frames .. warning:: In the near future, we intend to centralize PyTorch's video decoding capabilities within the `torchcodec `_ project. We encourage you to try it out and share your feedback, as the torchvision video decoders will eventually be deprecated. Args: filename (str): path to the video file. If using the pyav backend, this can be whatever ``av.open`` accepts. start_pts (int if pts_unit = 'pts', float / Fraction if pts_unit = 'sec', optional): The start presentation time of the video end_pts (int if pts_unit = 'pts', float / Fraction if pts_unit = 'sec', optional): The end presentation time pts_unit (str, optional): unit in which start_pts and end_pts values will be interpreted, either 'pts' or 'sec'. Defaults to 'pts'. output_format (str, optional): The format of the output video tensors. Can be either "THWC" (default) or "TCHW". Returns: vframes (Tensor[T, H, W, C] or Tensor[T, C, H, W]): the `T` video frames aframes (Tensor[K, L]): the audio frames, where `K` is the number of channels and `L` is the number of points info (Dict): metadata for the video and audio. Can contain the fields video_fps (float) and audio_fps (int) )rTCHWz5output_format should be either 'THWC' or 'TCHW', got .rget_video_backendpyavzFile not found: Nrpz7end_pts should be larger than start_pts, got start_pts=z and end_pts=ignoremetadata_errorsrr video_fpsrTr$cZg|](}|)Sr)to_rgb to_ndarrayrzrgs rr}zread_video..os,NNN 1133NNNrc6g|]}|Sr)rrs rr}zread_video..ps$EEEu((**EEEr)rr r rsr(r ro)r rrrsr )*rBrCrDrEr read_videoupper ValueError torchvisionrospathexists RuntimeErrorr _read_videorrJdefault_timebaserrMrSrTrrrr average_rater-rrGrKstackemptyrH concatenaterrrrrfloat32permute)rrrrlrrvframesrinfo video_framesraudio_timebaser_r vframes_list aframes_lists rrrsB 9 ! ! # #(EI,@,@,B,B(J'''!''))M,,,aQ^aaabbb------f$$w~~h'' ><(<<== =!+!7)WV^!_!_$$ ?EllG Y  k)kkbikk   #4 8<<< H $*J%.%6%7I,J"K"KLLG)'<GTTGGk& >>>G//!Q1-- GT !!s7+H&C H H&HH&!H"H&& H32H3cH|jdjj}|dSd|vrdSdS)NrFsLavcT)rSrr)r_rs r!_can_read_timestamps_from_packetsrs5!!$2.s!MMM!15;L;L;L;Lrr)rrc*g|]}|j |jSrrxrs rr}z,_decode_video_timestamps..s!NNN!AE`_ project. We encourage you to try it out and share your feedback, as the torchvision video decoders will eventually be deprecated. Note that the function decodes the whole video frame-by-frame. Args: filename (str): path to the video file pts_unit (str, optional): unit in which timestamp values will be returned either 'pts' or 'sec'. Defaults to 'pts'. Returns: pts (List[int] if pts_unit = 'pts', List[Fraction] if pts_unit = 'sec'): presentation timestamps for each one of the frames in the video. video_fps (float, optional): the frame rate for the video rrrNrrz Failed decoding frames for file zFailed to open container for z; Caught error: rocg|]}|zSrr)rzrvideo_time_bases rr}z)read_video_timestamps..s000qq?"000r)rBrCrDrEr read_video_timestampsrrr _read_video_timestampsrrrMrSrrrrrrrrJrRuntimeWarningsort) rrlrrryr_ video_streamemsgrs @rrrs2 9 ! ! # #3EI,@,@,B,B31222------f$$08DDDI C + WXx 8 8 8 =I & =(06q9 "."8Q29==CC"QQQM"OX"O"OPPPPPQ!,";<<  = = = = = = = = = = = = = = = +++KhKKKK c>********+HHJJJ50000C000  >s`D.$&D" CD"!C?<D">C??D" D."D&&D.)D&*D.. E$8"EE$)rN)rNNNNN)rNryrrx)1rrrrr fractionsrtypingrrrrrr rIrKrButilsr r rlogging set_levelERRORhasattrrrrgr\ ImportErrorrAttributeErrorAVErrorrboolrrrstrTensorrJrFrrrrrrrrrrrs :::::::::::::::: '''''' IIIJ)*** 728>,k : :  [   !n !!!j !    BBB )t))))  !(,*.!%!%.2u"u"u"u" u" u" d38n % u" %,' u"u"#u"DcN+u" u"u"u"u"pM'MMM M  M c8E#uS#XS *I$JKKL M MMMM` # \ #)-.>)? #LO #Z_ # \ # # # #")*04 v"v"v"UH_%v"eE8O, -v" v"  v"  5<tCH~ 56 v"v"v"v"r1IdO(@OT#YOOOO88C8385cT\]bTcIcCd888888s7AB*BB* B&#B*%B&&B**B=<B=