some models such as ssd, yolo have more than one output.

the clean up code in this patch is a little complex, it is because
that set_input_output_tf could be called for many times together
with ff_dnn_execute_model_tf, we have to clean resources for the
case that the two interfaces are called interleaved.
Signed-off-by: Guo, Yejun <yejun.guo@intel.com>
Signed-off-by: NPedro Arthur <bygrandao@gmail.com>

use TF_Tensor's cpu address to avoid extra memcpy.
Signed-off-by: Guo, Yejun <yejun.guo@intel.com>
Signed-off-by: NPedro Arthur <bygrandao@gmail.com>

Currently, within interface set_input_output, the dims/memory of the tensorflow
dnn model output is determined by executing the model with zero input,
actually, the output dims might vary with different input data for networks
such as object detection models faster-rcnn, ssd and yolo.

This patch moves the logic from set_input_output to execute_model which
is suitable for all the cases. Since interface changed, and so dnn_backend_native
also changes.

In vf_sr.c, it knows it's srcnn or espcn by executing the model with zero input,
so execute_model has to be called in function config_props
Signed-off-by: Guo, Yejun <yejun.guo@intel.com>
Signed-off-by: NPedro Arthur <bygrandao@gmail.com>

remove the requirment that the name of DNN model input/output
should be "x"/"y",
Signed-off-by: Guo, Yejun <yejun.guo@intel.com>
Signed-off-by: NPedro Arthur <bygrandao@gmail.com>

remove 'else' since there is always 'return' in 'if' scope,
so the code will be clean for later maintenance
Signed-off-by: Guo, Yejun <yejun.guo@intel.com>
Signed-off-by: NPedro Arthur <bygrandao@gmail.com>

otherwise, the following check will return error if layer_add_res
is randomly initialized.
Signed-off-by: Guo, Yejun <yejun.guo@intel.com>
Signed-off-by: NPedro Arthur <bygrandao@gmail.com>

Adding the support to build FFMPEG with HW accelerated decode and encode on PPC64
little endian architecture.
Signed-off-by: NTimo Rothenpieler <timo@rothenpieler.org>

Cuvid supports clips with a limit on maximum number of macroblocks.
This check was missing after cuvidGetDecoderCaps API call allowing
unsupported clips to proceed.
Added the missing check, same as the one in hwaccel nvdec implementation.
Signed-off-by: NTimo Rothenpieler <timo@rothenpieler.org>

Instead of doing each column one by one, doing several columns
together gives about 30% better performance.
Reviewed-by: NPaul B Mahol <onemda@gmail.com>
Signed-off-by: NRuiling Song <ruiling.song@intel.com>

It's no longer needed.
Signed-off-by: NJames Almer <jamrial@gmail.com>

Signed-off-by: NJames Almer <jamrial@gmail.com>

missing layout

./ffmpeg -loop 1 -s 1200x1440 -i tux16.png \
    -s 2400x720 -f rawvideo -y -vframes 5 -pix_fmt yuv420p16le -nostats test.raw

./ffmpeg -loop 1 -s 1200x1440 -i tux16.png \
    -s 2400x720 -f rawvideo -y -vframes 5 -pix_fmt yuv420p -nostats test.raw

32-bit mul, power8 only

2x speedup for hScale8To19_vsx (x86 SSE2 is 2.37):
  30896 UNITS in hscale,    8192 runs,      0 skips
  63956 UNITS in hscale,    8192 runs,      0 skips

2.06 for hScale16To15_vsx:
  30531 UNITS in hscale,    8192 runs,      0 skips
  63161 UNITS in hscale,    8192 runs,      0 skips

./ffmpeg -f lavfi -i yuvtestsrc=duration=1:size=1200x1440 \
    -s 2400x720 -f rawvideo -y -vframes 5 -pix_fmt yuv420p16le -nostats test.raw

2.26 speedup (x86 SSE2 is 2.32):
  23772 UNITS in hscale,    4096 runs,      0 skips
  53862 UNITS in hscale,    4096 runs,      0 skips

Currently profile mapping is hard-coded, and not flexible to do extactly
map (E.g: libmfx treats H264 constrained baseline to be baseline profile).

vaapi profile mapping funtion provides a better soultion than current
qsv mapping.
Signed-off-by: NZhong Li <zhong.li@intel.com>

libmfx hevc level defination is same as h264, not level_idc of SPEC.
Signed-off-by: NZhong Li <zhong.li@intel.com>

It is helpful to know why some clips decoding failed.
Ticket#7330 is a good example, with this patch it is easily to
know bitstream codec level is out of support range.
Signed-off-by: NZhong Li <zhong.li@intel.com>

Reference: Table 8: Interpretation of valid BITPIX value from FITS standard 4.0
Fixes: runtime error: division by zero
Fixes: 14581/clusterfuzz-testcase-minimized-ffmpeg_AV_CODEC_ID_FITS_fuzzer-5652382425284608

Found-by: continuous fuzzing process https://github.com/google/oss-fuzz/tree/master/projects/ffmpegReviewed-by: NPaul B Mahol <onemda@gmail.com>
Signed-off-by: NMichael Niedermayer <michael@niedermayer.cc>

Signed-off-by: NJames Almer <jamrial@gmail.com>

10 bytes (id3v2 header amount of bytes) were being read before any checks
were made on the bitstream. The result was that we were overreading into
the next frame if the current one was 8 or 9 bytes long.

Fixes tickets #7271 and #7869.
Signed-off-by: NJames Almer <jamrial@gmail.com>

The latest generation video decoder on the Turing chips supports
decoding HEVC 4:4:4. This change adds AV_PIX_FMT_VDPAU as a valid format
for HEVC 4:4:4 8 bit.

Pass SPS, PPS range extensions to VDPAU layer via
VdpPictureInfoHEVC444. Added VdpPictureInfoHEVC444 struct to
VdpPictureInfo union to populate the range extension params. Mapped
FF_PROFILE_HEVC_REXT to VDP_DECODER_PROFILE_HEVC_MAIN_444.

New VdpYCbCr Formats VDP_YCBCR_FORMAT_Y_U_V_444 and,
VDP_YCBCR_FORMAT_Y_UV_444 have been added in VDPAU with libvdpau-1.2
to be used in get/putbits for YUV 4:4:4 surfaces. Earlier mapping of
AV_PIX_FMT_YUV444P to VDP_YCBCR_FORMAT_YV12 is not valid.

Hence this Change maps AV_PIX_FMT_YUV444P to VDP_YCBCR_FORMAT_Y_U_V_444
to access the YUV 4:4:4 surface via read-back API's of VDPAU.

Fixes ticket #7859
Reported and tested by Steffen Ohrendorf, steffen d ohrendorf a gmx d de

avfilter/vf_freezedetect: fix missing freeze_start when the freeze length is around the detection duration

Fixes ticket #7875.
Signed-off-by: NMarton Balint <cus@passwd.hu>

Apparently in the new SDK one cannot query if VANC output is supported, so we
will fall back to non-VANC output if enabling the video output with VANC fails.

Fixes ticket #7867.
Signed-off-by: NMarton Balint <cus@passwd.hu>

Reviewed-by: NPaul B Mahol <onemda@gmail.com>
Signed-off-by: NJames Almer <jamrial@gmail.com>

avcodec/h263dec: Fixed VA API, VDPAU, and VideoToolbox hardware acceleration due to missing `hw_configs` property.

Bug originally introduced in commit 758fbc54.

Allows running fate from out-of-tree builds on wsl.

Fixes: Timeout (11sec -> 5sec)
Fixes: 14473/clusterfuzz-testcase-minimized-ffmpeg_AV_CODEC_ID_JV_fuzzer-5761630857592832

Found-by: continuous fuzzing process https://github.com/google/oss-fuzz/tree/master/projects/ffmpegReviewed-by: NPeter Ross <pross@xvid.org>
Signed-off-by: NMichael Niedermayer <michael@niedermayer.cc>

Fixes: Assertion failure
Fixes: 14484/clusterfuzz-testcase-minimized-ffmpeg_AV_CODEC_ID_PGMYUV_fuzzer-5150016408125440

Found-by: continuous fuzzing process https://github.com/google/oss-fuzz/tree/master/projects/ffmpegReviewed-by: NPaul B Mahol <onemda@gmail.com>
Signed-off-by: NMichael Niedermayer <michael@niedermayer.cc>