Skip to content

5.x merge 4.x #3885

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Merged
merged 15 commits into from
Feb 24, 2025
Merged
Show file tree
Hide file tree
Changes from all commits
Commits
File filter

Filter by extension

Filter by extension

Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
3 changes: 2 additions & 1 deletion modules/cudacodec/include/opencv2/cudacodec.hpp
Original file line number Diff line number Diff line change
Expand Up @@ -186,7 +186,7 @@ struct CV_EXPORTS_W_SIMPLE EncoderParams
public:
CV_WRAP EncoderParams() : nvPreset(ENC_PRESET_P3), tuningInfo(ENC_TUNING_INFO_HIGH_QUALITY), encodingProfile(ENC_CODEC_PROFILE_AUTOSELECT),
rateControlMode(ENC_PARAMS_RC_VBR), multiPassEncoding(ENC_MULTI_PASS_DISABLED), constQp({ 0,0,0 }), averageBitRate(0), maxBitRate(0),
targetQuality(30), gopLength(250), idrPeriod(250) {};
targetQuality(30), gopLength(250), idrPeriod(250), videoFullRangeFlag(false){};
CV_PROP_RW EncodePreset nvPreset;
CV_PROP_RW EncodeTuningInfo tuningInfo;
CV_PROP_RW EncodeProfile encodingProfile;
Expand All @@ -198,6 +198,7 @@ struct CV_EXPORTS_W_SIMPLE EncoderParams
CV_PROP_RW uint8_t targetQuality; //!< value 0 - 51 where video quality decreases as targetQuality increases, used with \ref ENC_PARAMS_RC_VBR.
CV_PROP_RW int gopLength; //!< the number of pictures in one GOP, ensuring \ref idrPeriod >= \ref gopLength.
CV_PROP_RW int idrPeriod; //!< IDR interval, ensuring \ref idrPeriod >= \ref gopLength.
CV_PROP_RW bool videoFullRangeFlag;//!< Indicates if the black level, luma and chroma of the source are represented using the full or limited range (AKA TV or "analogue" range) of values as defined in Annex E of the ITU-T Specification.
};
CV_EXPORTS bool operator==(const EncoderParams& lhs, const EncoderParams& rhs);

Expand Down
8 changes: 4 additions & 4 deletions modules/cudacodec/src/NvEncoder.cpp
Original file line number Diff line number Diff line change
Expand Up @@ -101,7 +101,7 @@ void NvEncoder::CreateDefaultEncoderParams(NV_ENC_INITIALIZE_PARAMS* pIntializeP
#endif
pIntializeParams->tuningInfo = tuningInfo;
pIntializeParams->encodeConfig->rcParams.rateControlMode = NV_ENC_PARAMS_RC_CONSTQP;
#if (NVENCAPI_MAJOR_VERSION >= 12 && NVENCAPI_MINOR_VERSION >= 2)
#if ((NVENCAPI_MAJOR_VERSION == 12 && NVENCAPI_MINOR_VERSION >= 2) || NVENCAPI_MAJOR_VERSION > 12)
NV_ENC_PRESET_CONFIG presetConfig = { NV_ENC_PRESET_CONFIG_VER, 0, { NV_ENC_CONFIG_VER } };
#else
NV_ENC_PRESET_CONFIG presetConfig = { NV_ENC_PRESET_CONFIG_VER, { NV_ENC_CONFIG_VER } };
Expand All @@ -119,7 +119,7 @@ void NvEncoder::CreateDefaultEncoderParams(NV_ENC_INITIALIZE_PARAMS* pIntializeP
}
else if (pIntializeParams->encodeGUID == NV_ENC_CODEC_HEVC_GUID)
{
#if (NVENCAPI_MAJOR_VERSION >= 12 && NVENCAPI_MINOR_VERSION >= 2)
#if ((NVENCAPI_MAJOR_VERSION == 12 && NVENCAPI_MINOR_VERSION >= 2) || NVENCAPI_MAJOR_VERSION > 12)
pIntializeParams->encodeConfig->encodeCodecConfig.hevcConfig.inputBitDepth = pIntializeParams->encodeConfig->encodeCodecConfig.hevcConfig.outputBitDepth =
(m_eBufferFormat == NV_ENC_BUFFER_FORMAT_YUV420_10BIT || m_eBufferFormat == NV_ENC_BUFFER_FORMAT_YUV444_10BIT) ? NV_ENC_BIT_DEPTH_10 : NV_ENC_BIT_DEPTH_8;
#else
Expand Down Expand Up @@ -179,7 +179,7 @@ void NvEncoder::CreateEncoder(const NV_ENC_INITIALIZE_PARAMS* pEncoderParams)
if (pEncoderParams->encodeGUID == NV_ENC_CODEC_HEVC_GUID)
{
bool yuv10BitFormat = (m_eBufferFormat == NV_ENC_BUFFER_FORMAT_YUV420_10BIT || m_eBufferFormat == NV_ENC_BUFFER_FORMAT_YUV444_10BIT) ? true : false;
#if (NVENCAPI_MAJOR_VERSION >= 12 && NVENCAPI_MINOR_VERSION >= 2)
#if ((NVENCAPI_MAJOR_VERSION == 12 && NVENCAPI_MINOR_VERSION >= 2) || NVENCAPI_MAJOR_VERSION > 12)
if (yuv10BitFormat && pEncoderParams->encodeConfig->encodeCodecConfig.hevcConfig.inputBitDepth != NV_ENC_BIT_DEPTH_10)
#else
if (yuv10BitFormat && pEncoderParams->encodeConfig->encodeCodecConfig.hevcConfig.pixelBitDepthMinus8 != 2)
Expand All @@ -205,7 +205,7 @@ void NvEncoder::CreateEncoder(const NV_ENC_INITIALIZE_PARAMS* pEncoderParams)
}
else
{
#if (NVENCAPI_MAJOR_VERSION >= 12 && NVENCAPI_MINOR_VERSION >= 2)
#if ((NVENCAPI_MAJOR_VERSION == 12 && NVENCAPI_MINOR_VERSION >= 2) || NVENCAPI_MAJOR_VERSION > 12)
NV_ENC_PRESET_CONFIG presetConfig = { NV_ENC_PRESET_CONFIG_VER, 0, { NV_ENC_CONFIG_VER } };
#else
NV_ENC_PRESET_CONFIG presetConfig = { NV_ENC_PRESET_CONFIG_VER, { NV_ENC_CONFIG_VER } };
Expand Down
35 changes: 24 additions & 11 deletions modules/cudacodec/src/video_writer.cpp
Original file line number Diff line number Diff line change
Expand Up @@ -53,10 +53,6 @@ Ptr<cudacodec::VideoWriter> createVideoWriter(const String&, const Size, const C

#else // !defined HAVE_NVCUVENC

#if defined(WIN32) // remove when FFmpeg wrapper includes PR25874
#define WIN32_WAIT_FOR_FFMPEG_WRAPPER_UPDATE
#endif

NV_ENC_BUFFER_FORMAT EncBufferFormat(const ColorFormat colorFormat);
int NChannels(const ColorFormat colorFormat);
GUID CodecGuid(const Codec codec);
Expand Down Expand Up @@ -86,7 +82,7 @@ class FFmpegVideoWriter : public EncoderCallback
FFmpegVideoWriter::FFmpegVideoWriter(const String& fileName, const Codec codec, const int fps, const Size sz, const int idrPeriod) {
if (!videoio_registry::hasBackend(CAP_FFMPEG))
CV_Error(Error::StsNotImplemented, "FFmpeg backend not found");
const int fourcc = codec == Codec::H264 ? cv::VideoWriter::fourcc('a', 'v', 'c', '1') : cv::VideoWriter::fourcc('h', 'e', 'v', '1');
const int fourcc = codec == Codec::H264 ? cv::VideoWriter::fourcc('a', 'v', 'c', '1') : cv::VideoWriter::fourcc('h', 'v', 'c', '1');
writer.open(fileName, fourcc, fps, sz, { VideoWriterProperties::VIDEOWRITER_PROP_RAW_VIDEO, 1, VideoWriterProperties::VIDEOWRITER_PROP_KEY_INTERVAL, idrPeriod });
if (!writer.isOpened())
CV_Error(Error::StsUnsupportedFormat, "Unsupported video sink");
Expand All @@ -107,9 +103,7 @@ void FFmpegVideoWriter::onEncoded(const std::vector<std::vector<uint8_t>>& vPack
Mat wrappedPacket(1, packet.size(), CV_8UC1, (void*)packet.data());
const double ptsDouble = static_cast<double>(pts.at(i));
CV_Assert(static_cast<uint64_t>(ptsDouble) == pts.at(i));
#if !defined(WIN32_WAIT_FOR_FFMPEG_WRAPPER_UPDATE)
CV_Assert(writer.set(VIDEOWRITER_PROP_PTS, ptsDouble));
#endif
writer.write(wrappedPacket);
}
}
Expand Down Expand Up @@ -321,6 +315,13 @@ GUID EncodingPresetGuid(const EncodePreset nvPreset) {
CV_Error(Error::StsUnsupportedFormat, msg);
}

std::string GetVideoCodecString(const GUID codec) {
if (codec == NV_ENC_CODEC_H264_GUID) return "AVC/H.264";
else if (codec == NV_ENC_CODEC_HEVC_GUID) return "H.265/HEVC";
else if (codec == NV_ENC_CODEC_AV1_GUID) return "AV1";
else return "Unknown";
}

void VideoWriterImpl::InitializeEncoder(const GUID codec, const double fps)
{
NV_ENC_INITIALIZE_PARAMS initializeParams = {};
Expand All @@ -337,15 +338,27 @@ void VideoWriterImpl::InitializeEncoder(const GUID codec, const double fps)
initializeParams.encodeConfig->rcParams.maxBitRate = encoderParams.maxBitRate;
initializeParams.encodeConfig->rcParams.targetQuality = encoderParams.targetQuality;
initializeParams.encodeConfig->gopLength = encoderParams.gopLength;
#if !defined(WIN32_WAIT_FOR_FFMPEG_WRAPPER_UPDATE)
if (initializeParams.encodeConfig->frameIntervalP > 1) {
CV_Assert(encoderCallback->setFrameIntervalP(initializeParams.encodeConfig->frameIntervalP));
}
#endif
if (codec == NV_ENC_CODEC_H264_GUID)
if (codec == NV_ENC_CODEC_H264_GUID) {
initializeParams.encodeConfig->encodeCodecConfig.h264Config.idrPeriod = encoderParams.idrPeriod;
else if (codec == NV_ENC_CODEC_HEVC_GUID)
if (encoderParams.videoFullRangeFlag) {
initializeParams.encodeConfig->encodeCodecConfig.h264Config.h264VUIParameters.videoFullRangeFlag = 1;
initializeParams.encodeConfig->encodeCodecConfig.h264Config.h264VUIParameters.videoSignalTypePresentFlag = 1;
}
}
else if (codec == NV_ENC_CODEC_HEVC_GUID) {
initializeParams.encodeConfig->encodeCodecConfig.hevcConfig.idrPeriod = encoderParams.idrPeriod;
if (encoderParams.videoFullRangeFlag) {
initializeParams.encodeConfig->encodeCodecConfig.hevcConfig.hevcVUIParameters.videoFullRangeFlag = 1;
initializeParams.encodeConfig->encodeCodecConfig.hevcConfig.hevcVUIParameters.videoSignalTypePresentFlag = 1;
}
}
else {
std::string msg = "videoFullRangeFlag is not supported by codec: " + GetVideoCodecString(codec);
CV_LOG_WARNING(NULL, msg);
}
pEnc->CreateEncoder(&initializeParams);
}

Expand Down
18 changes: 5 additions & 13 deletions modules/cudacodec/test/test_video.cpp
Original file line number Diff line number Diff line change
Expand Up @@ -1011,10 +1011,6 @@ struct H264ToH265 : SetDevice
{
};

#if defined(WIN32) // remove when FFmpeg wrapper includes PR25874
#define WIN32_WAIT_FOR_FFMPEG_WRAPPER_UPDATE
#endif

CUDA_TEST_P(H264ToH265, Transcode)
{
const std::string inputFile = std::string(cvtest::TS::ptr()->get_data_path()) + "../highgui/video/big_buck_bunny.h264";
Expand Down Expand Up @@ -1054,10 +1050,8 @@ CUDA_TEST_P(H264ToH265, Transcode)
for (int i = 0; i < nFrames; ++i) {
cap >> frame;
ASSERT_FALSE(frame.empty());
#if !defined(WIN32_WAIT_FOR_FFMPEG_WRAPPER_UPDATE)
const int pts = static_cast<int>(cap.get(CAP_PROP_PTS));
ASSERT_EQ(i, pts > 0 ? pts : 0); // FFmpeg back end returns dts if pts is zero.
#endif
}
}
ASSERT_EQ(0, remove(outputFile.c_str()));
Expand All @@ -1066,7 +1060,7 @@ CUDA_TEST_P(H264ToH265, Transcode)
INSTANTIATE_TEST_CASE_P(CUDA_Codec, H264ToH265, ALL_DEVICES);

CV_ENUM(YuvColorFormats, cudacodec::ColorFormat::NV_YUV444, cudacodec::ColorFormat::NV_YUV420_10BIT, cudacodec::ColorFormat::NV_YUV444_10BIT)
PARAM_TEST_CASE(YUVFormats, cv::cuda::DeviceInfo, YuvColorFormats)
PARAM_TEST_CASE(YUVFormats, cv::cuda::DeviceInfo, YuvColorFormats, bool)
{
};

Expand All @@ -1075,6 +1069,7 @@ CUDA_TEST_P(YUVFormats, Transcode)
cv::cuda::setDevice(GET_PARAM(0).deviceID());
const std::string inputFile = std::string(cvtest::TS::ptr()->get_data_path()) + "../highgui/video/big_buck_bunny.h265";
const cv::cudacodec::ColorFormat writerColorFormat = static_cast<cudacodec::ColorFormat>(static_cast<int>(GET_PARAM(1)));
const bool fullRange = GET_PARAM(2);
constexpr double fps = 25;
const cudacodec::Codec codec = cudacodec::Codec::HEVC;
const std::string ext = ".mp4";
Expand All @@ -1088,6 +1083,7 @@ CUDA_TEST_P(YUVFormats, Transcode)
cv::cudacodec::EncoderParams params;
params.tuningInfo = cv::cudacodec::EncodeTuningInfo::ENC_TUNING_INFO_LOSSLESS;
params.rateControlMode = cv::cudacodec::EncodeParamsRcMode::ENC_PARAMS_RC_CONSTQP;
params.videoFullRangeFlag = fullRange;
for (int i = 0; i < nFrames; ++i) {
ASSERT_TRUE(cap.read(frame));
ASSERT_FALSE(frame.empty());
Expand All @@ -1101,7 +1097,7 @@ CUDA_TEST_P(YUVFormats, Transcode)
yuvFormat = cudacodec::SurfaceFormat::SF_P016;
bitDepth = cudacodec::BitDepth::SIXTEEN;
}
generateTestImages(frame, yuv, bgr, yuvFormat, cudacodec::ColorFormat::BGR, bitDepth, false);
generateTestImages(frame, yuv, bgr, yuvFormat, cudacodec::ColorFormat::BGR, bitDepth, false, fullRange);
bgrGs.push_back(bgr.clone());
if (writer.empty())
writer = cv::cudacodec::createVideoWriter(outputFile, frame.size(), codec, fps, writerColorFormat, params);
Expand All @@ -1125,7 +1121,7 @@ CUDA_TEST_P(YUVFormats, Transcode)
ASSERT_EQ(0, remove(outputFile.c_str()));
}

INSTANTIATE_TEST_CASE_P(CUDA_Codec, YUVFormats, testing::Combine(ALL_DEVICES, YuvColorFormats::all()));
INSTANTIATE_TEST_CASE_P(CUDA_Codec, YUVFormats, testing::Combine(ALL_DEVICES, YuvColorFormats::all(), testing::Bool()));
#endif

#if defined(HAVE_NVCUVENC)
Expand Down Expand Up @@ -1201,10 +1197,8 @@ CUDA_TEST_P(Write, Writer)
for (int i = 0; i < nFrames; ++i) {
cap >> frame;
ASSERT_FALSE(frame.empty());
#if !defined(WIN32_WAIT_FOR_FFMPEG_WRAPPER_UPDATE)
const int pts = static_cast<int>(cap.get(CAP_PROP_PTS));
ASSERT_EQ(i, pts > 0 ? pts : 0); // FFmpeg back end returns dts if pts is zero.
#endif
}
}
ASSERT_EQ(0, remove(outputFile.c_str()));
Expand Down Expand Up @@ -1299,10 +1293,8 @@ CUDA_TEST_P(EncoderParams, Writer)
const bool keyFrameActual = capRaw.get(CAP_PROP_LRF_HAS_KEY_FRAME) == 1.0;
const bool keyFrameReference = i % idrPeriod == 0;
ASSERT_EQ(keyFrameActual, keyFrameReference);
#if !defined(WIN32_WAIT_FOR_FFMPEG_WRAPPER_UPDATE)
const int pts = static_cast<int>(cap.get(CAP_PROP_PTS));
ASSERT_EQ(i, pts > 0 ? pts : 0); // FFmpeg back end returns dts if pts is zero.
#endif
}
}
}
Expand Down
2 changes: 1 addition & 1 deletion modules/cudalegacy/include/opencv2/cudalegacy/private.hpp
Original file line number Diff line number Diff line change
Expand Up @@ -91,6 +91,6 @@ namespace cv { namespace cuda

//! @endcond

#endif HAVE_CUDA
#endif // HAVE_CUDA

#endif // OPENCV_CORE_CUDALEGACY_PRIVATE_HPP
3 changes: 2 additions & 1 deletion modules/ml/CMakeLists.txt
Original file line number Diff line number Diff line change
Expand Up @@ -7,4 +7,5 @@ ocv_create_module()

ocv_add_accuracy_tests()
ocv_add_perf_tests()
ocv_add_samples(opencv_imgproc opencv_objdetect opencv_video)
# dnn is transitive dependency pf opencv_video
ocv_add_samples(opencv_imgproc opencv_objdetect opencv_video opencv_dnn)
Loading