Merge pull request #3542 from cudawarped:cudacodec_videoreader_seek

cudacodec::VideoReader: allow frame seek on initialization #3542

Allow seeking of video source on initialization of `cudacodec::VideoReader` when new variable `VideoReaderInitParams::iFirstFrame` != 0.

Dependant on opencv/opencv#24012

Fixes #3541.

### Pull Request Readiness Checklist

See details at https://github.com/opencv/opencv/wiki/How_to_contribute#making-a-good-pull-request

- [x] I agree to contribute to the project under Apache 2 License.
- [x] To the best of my knowledge, the proposed patch is not based on a code under GPL or another license that is incompatible with OpenCV
- [x] The PR is proposed to the proper branch
- [x] There is a reference to the original bug report and related work
- [x] There is accuracy test, performance test and test data in opencv_extra repository, if applicable
      Patch to opencv_extra has the same branch name.
- [x] The feature is well documented and sample code can be built with the project CMake

Author: cudawarped

modules/cudacodec/include/opencv2/cudacodec.hpp

@@ -544,6 +544,14 @@ class CV_EXPORTS_W RawVideoSource
     @return `true` unless the property is unset set or not supported.
      */
     virtual bool get(const int propertyId, double& propertyVal) const = 0;
+
+    /** @brief Retrieve the index of the first frame that will returned after construction.
+
+    @return index of the index of the first frame that will returned after construction.
+
+    @note To reduce the decoding overhead when initializing VideoReader to start its decoding from frame N, RawVideoSource should seek to the first valid key frame less than or equal to N and return that index here.
+     */
+    virtual int getFirstFrameIdx() const = 0;
 };
 
 /** @brief VideoReader initialization parameters
@@ -561,9 +569,10 @@ but it cannot go below the number determined by NVDEC.
 @param targetRoi Region of interest (x/width should be multiples of 4 and y/height multiples of 2) within the output frame to copy and resize the decoded frame to,
 defaults to the full frame.
 @param enableHistogram Request output of decoded luma histogram \a hist from VideoReader::nextFrame(GpuMat& frame, GpuMat& hist, Stream& stream), if hardware supported.
+@param firstFrameIdx Index of the first frame to seek to on initialization of the VideoReader.
 */
 struct CV_EXPORTS_W_SIMPLE VideoReaderInitParams {
-    CV_WRAP VideoReaderInitParams() : udpSource(false), allowFrameDrop(false), minNumDecodeSurfaces(0), rawMode(0), enableHistogram(false){};
+    CV_WRAP VideoReaderInitParams() : udpSource(false), allowFrameDrop(false), minNumDecodeSurfaces(0), rawMode(0), enableHistogram(false), firstFrameIdx(0){};
     CV_PROP_RW bool udpSource;
     CV_PROP_RW bool allowFrameDrop;
     CV_PROP_RW int minNumDecodeSurfaces;
@@ -572,6 +581,7 @@ struct CV_EXPORTS_W_SIMPLE VideoReaderInitParams {
     CV_PROP_RW cv::Rect srcRoi;
     CV_PROP_RW cv::Rect targetRoi;
     CV_PROP_RW bool enableHistogram;
+    CV_PROP_RW int firstFrameIdx;
 };
 
 /** @brief Creates video reader.

modules/cudacodec/src/ffmpeg_video_source.cpp

@@ -169,19 +169,21 @@ bool ParamSetsExist(unsigned char* parameterSets, const int szParameterSets, uns
     return paramSetStartCodeLen != 0 && packetStartCodeLen != 0 && parameterSets[paramSetStartCodeLen] == data[packetStartCodeLen];
 }
 
-cv::cudacodec::detail::FFmpegVideoSource::FFmpegVideoSource(const String& fname, const std::vector<int>& _videoCaptureParams)
+cv::cudacodec::detail::FFmpegVideoSource::FFmpegVideoSource(const String& fname, const std::vector<int>& _videoCaptureParams, const int iMaxStartFrame)
     : videoCaptureParams(_videoCaptureParams)
 {
     if (!videoio_registry::hasBackend(CAP_FFMPEG))
         CV_Error(Error::StsNotImplemented, "FFmpeg backend not found");
 
-    cap.open(fname, CAP_FFMPEG, videoCaptureParams);
-    if (!cap.isOpened())
+    videoCaptureParams.push_back(CAP_PROP_FORMAT);
+    videoCaptureParams.push_back(-1);
+    if (!cap.open(fname, CAP_FFMPEG, videoCaptureParams))
         CV_Error(Error::StsUnsupportedFormat, "Unsupported video source");
-
-    if (!cap.set(CAP_PROP_FORMAT, -1))  // turn off video decoder (extract stream)
-        CV_Error(Error::StsUnsupportedFormat, "Fetching of RAW video streams is not supported");
     CV_Assert(cap.get(CAP_PROP_FORMAT) == -1);
+    if (iMaxStartFrame) {
+        CV_Assert(cap.set(CAP_PROP_POS_FRAMES, iMaxStartFrame));
+        firstFrameIdx = static_cast<int>(cap.get(CAP_PROP_POS_FRAMES));
+    }
 
     const int codecExtradataIndex = static_cast<int>(cap.get(CAP_PROP_CODEC_EXTRADATA_INDEX));
     Mat tmpExtraData;

modules/cudacodec/src/ffmpeg_video_source.hpp

@@ -51,7 +51,7 @@ namespace cv { namespace cudacodec { namespace detail {
 class FFmpegVideoSource : public RawVideoSource
 {
 public:
-    FFmpegVideoSource(const String& fname, const std::vector<int>& params);
+    FFmpegVideoSource(const String& fname, const std::vector<int>& params, const int iMaxStartFrame);
     ~FFmpegVideoSource();
 
     bool getNextPacket(unsigned char** data, size_t* size) CV_OVERRIDE;
@@ -66,12 +66,15 @@ class FFmpegVideoSource : public RawVideoSource
 
     bool get(const int propertyId, double& propertyVal) const;
 
+    int getFirstFrameIdx() const { return firstFrameIdx; }
+
 private:
     FormatInfo format_;
     VideoCapture cap;
     Mat rawFrame, extraData, dataWithHeader;
     int iFrame = 0;
     std::vector<int> videoCaptureParams;
+    int firstFrameIdx = 0;
 };
 
 }}}

modules/cudacodec/src/video_reader.cpp

@@ -112,7 +112,7 @@ namespace
     {
     public:
         explicit VideoReaderImpl(const Ptr<VideoSource>& source, const int minNumDecodeSurfaces, const bool allowFrameDrop = false , const bool udpSource = false,
-            const Size targetSz = Size(), const Rect srcRoi = Rect(), const Rect targetRoi = Rect(), const bool enableHistogram = false);
+            const Size targetSz = Size(), const Rect srcRoi = Rect(), const Rect targetRoi = Rect(), const bool enableHistogram = false, const int firstFrameIdx = 0);
         ~VideoReaderImpl();
 
         bool nextFrame(GpuMat& frame, Stream& stream) CV_OVERRIDE;
@@ -135,6 +135,9 @@ namespace
         bool get(const int propertyId, double& propertyVal) const CV_OVERRIDE;
 
     private:
+        bool skipFrame();
+        bool aquireFrameInfo(std::pair<CUVIDPARSERDISPINFO, CUVIDPROCPARAMS>& frameInfo, Stream& stream = Stream::Null());
+        void releaseFrameInfo(const std::pair<CUVIDPARSERDISPINFO, CUVIDPROCPARAMS>& frameInfo);
         bool internalGrab(GpuMat & frame, GpuMat & histogram, Stream & stream);
         void waitForDecoderInit();
 
@@ -154,6 +157,7 @@ namespace
         static const int rawPacketsBaseIdx = 2;
         ColorFormat colorFormat = ColorFormat::BGRA;
         static const String errorMsg;
+        int iFrame = 0;
     };
 
     const String VideoReaderImpl::errorMsg = "Parsing/Decoding video source failed, check GPU memory is available and GPU supports requested functionality.";
@@ -173,7 +177,7 @@ namespace
     }
 
     VideoReaderImpl::VideoReaderImpl(const Ptr<VideoSource>& source, const int minNumDecodeSurfaces, const bool allowFrameDrop, const bool udpSource,
-        const Size targetSz, const Rect srcRoi, const Rect targetRoi, const bool enableHistogram) :
+        const Size targetSz, const Rect srcRoi, const Rect targetRoi, const bool enableHistogram, const int firstFrameIdx) :
         videoSource_(source),
         lock_(0)
     {
@@ -190,6 +194,8 @@ namespace
         videoSource_->setVideoParser(videoParser_);
         videoSource_->start();
         waitForDecoderInit();
+        for(iFrame = videoSource_->getFirstFrameIdx(); iFrame < firstFrameIdx; iFrame++)
+            CV_Assert(skipFrame());
         videoSource_->updateFormat(videoDecoder_->format());
     }
 
@@ -209,10 +215,7 @@ namespace
         CUvideoctxlock m_lock;
     };
 
-    bool VideoReaderImpl::internalGrab(GpuMat& frame, GpuMat& histogram, Stream& stream) {
-        if (videoParser_->hasError())
-            CV_Error(Error::StsError, errorMsg);
-        cudacodec::FormatInfo fmt;
+    bool VideoReaderImpl::aquireFrameInfo(std::pair<CUVIDPARSERDISPINFO, CUVIDPROCPARAMS>& frameInfo, Stream& stream) {
         if (frames_.empty())
         {
             CUVIDPARSERDISPINFO displayInfo;
@@ -234,34 +237,53 @@ namespace
 
             bool isProgressive = displayInfo.progressive_frame != 0;
             const int num_fields = isProgressive ? 1 : 2 + displayInfo.repeat_first_field;
-            fmt = videoDecoder_->format();
-            videoSource_->updateFormat(fmt);
 
             for (int active_field = 0; active_field < num_fields; ++active_field)
             {
                 CUVIDPROCPARAMS videoProcParams;
                 std::memset(&videoProcParams, 0, sizeof(CUVIDPROCPARAMS));
 
                 videoProcParams.progressive_frame = displayInfo.progressive_frame;
-                videoProcParams.second_field      = active_field;
-                videoProcParams.top_field_first   = displayInfo.top_field_first;
-                videoProcParams.unpaired_field    = (num_fields == 1);
+                videoProcParams.second_field = active_field;
+                videoProcParams.top_field_first = displayInfo.top_field_first;
+                videoProcParams.unpaired_field = (num_fields == 1);
                 videoProcParams.output_stream = StreamAccessor::getStream(stream);
 
                 frames_.push_back(std::make_pair(displayInfo, videoProcParams));
             }
         }
+        else {
+            for (auto& frame : frames_)
+                frame.second.output_stream = StreamAccessor::getStream(stream);
+        }
 
         if (frames_.empty())
             return false;
 
-        std::pair<CUVIDPARSERDISPINFO, CUVIDPROCPARAMS> frameInfo = frames_.front();
+        frameInfo = frames_.front();
         frames_.pop_front();
+        return true;
+    }
+
+    void VideoReaderImpl::releaseFrameInfo(const std::pair<CUVIDPARSERDISPINFO, CUVIDPROCPARAMS>& frameInfo) {
+        // release the frame, so it can be re-used in decoder
+        if (frames_.empty())
+            frameQueue_->releaseFrame(frameInfo.first);
+    }
+
+    bool VideoReaderImpl::internalGrab(GpuMat& frame, GpuMat& histogram, Stream& stream) {
+        if (videoParser_->hasError())
+            CV_Error(Error::StsError, errorMsg);
+
+        std::pair<CUVIDPARSERDISPINFO, CUVIDPROCPARAMS> frameInfo;
+        if (!aquireFrameInfo(frameInfo, stream))
+            return false;
 
         {
             VideoCtxAutoLock autoLock(lock_);
 
             unsigned long long cuHistogramPtr = 0;
+            const cudacodec::FormatInfo fmt = videoDecoder_->format();
             if (fmt.enableHistogram)
                 frameInfo.second.histogram_dptr = &cuHistogramPtr;
 
@@ -281,10 +303,16 @@ namespace
             videoDecoder_->unmapFrame(decodedFrame);
         }
 
-        // release the frame, so it can be re-used in decoder
-        if (frames_.empty())
-            frameQueue_->releaseFrame(frameInfo.first);
+        releaseFrameInfo(frameInfo);
+        iFrame++;
+        return true;
+    }
 
+    bool VideoReaderImpl::skipFrame() {
+        std::pair<CUVIDPARSERDISPINFO, CUVIDPROCPARAMS> frameInfo;
+        if (!aquireFrameInfo(frameInfo))
+            return false;
+        releaseFrameInfo(frameInfo);
         return true;
     }
 
@@ -399,6 +427,10 @@ namespace
     }
 
     bool VideoReaderImpl::get(const int propertyId, double& propertyVal) const {
+        if (propertyId == cv::VideoCaptureProperties::CAP_PROP_POS_FRAMES) {
+            propertyVal = static_cast<double>(iFrame);
+            return true;
+        }
         return videoSource_->get(propertyId, propertyVal);
     }
 
@@ -421,28 +453,26 @@ Ptr<VideoReader> cv::cudacodec::createVideoReader(const String& filename, const
     CV_Assert(!filename.empty());
 
     Ptr<VideoSource> videoSource;
-
     try
     {
         // prefer ffmpeg to cuvidGetSourceVideoFormat() which doesn't always return the corrct raw pixel format
-        Ptr<RawVideoSource> source(new FFmpegVideoSource(filename, sourceParams));
+        Ptr<RawVideoSource> source(new FFmpegVideoSource(filename, sourceParams, params.firstFrameIdx));
         videoSource.reset(new RawVideoSourceWrapper(source, params.rawMode));
     }
     catch (...)
     {
         if (sourceParams.size()) throw;
         videoSource.reset(new CuvidVideoSource(filename));
     }
-
     return makePtr<VideoReaderImpl>(videoSource, params.minNumDecodeSurfaces, params.allowFrameDrop, params.udpSource, params.targetSz,
-        params.srcRoi, params.targetRoi, params.enableHistogram);
+        params.srcRoi, params.targetRoi, params.enableHistogram, params.firstFrameIdx);
 }
 
 Ptr<VideoReader> cv::cudacodec::createVideoReader(const Ptr<RawVideoSource>& source, const VideoReaderInitParams params)
 {
     Ptr<VideoSource> videoSource(new RawVideoSourceWrapper(source, params.rawMode));
     return makePtr<VideoReaderImpl>(videoSource, params.minNumDecodeSurfaces, params.allowFrameDrop, params.udpSource, params.targetSz,
-        params.srcRoi, params.targetRoi, params.enableHistogram);
+        params.srcRoi, params.targetRoi, params.enableHistogram, params.firstFrameIdx);
 }
 
 void cv::cudacodec::MapHist(const GpuMat& hist, Mat& histFull) {

modules/cudacodec/src/video_source.cpp

@@ -76,6 +76,10 @@ bool cv::cudacodec::detail::RawVideoSourceWrapper::get(const int propertyId, dou
     return source_->get(propertyId, propertyVal);
 }
 
+int cv::cudacodec::detail::RawVideoSourceWrapper::getFirstFrameIdx() const {
+    return source_->getFirstFrameIdx();
+}
+
 void cv::cudacodec::detail::RawVideoSourceWrapper::start()
 {
     stop_ = false;

modules/cudacodec/src/video_source.hpp

@@ -58,6 +58,7 @@ class VideoSource
     virtual FormatInfo format() const = 0;
     virtual void updateFormat(const FormatInfo& videoFormat) = 0;
     virtual bool get(const int propertyId, double& propertyVal) const { return false; }
+    virtual int getFirstFrameIdx() const { return 0; }
     virtual void start() = 0;
     virtual void stop() = 0;
     virtual bool isStarted() const = 0;
@@ -91,6 +92,7 @@ class RawVideoSourceWrapper : public VideoSource
     FormatInfo format() const CV_OVERRIDE;
     void updateFormat(const FormatInfo& videoFormat) CV_OVERRIDE;
     bool get(const int propertyId, double& propertyVal) const CV_OVERRIDE;
+    int getFirstFrameIdx() const CV_OVERRIDE;
     void start() CV_OVERRIDE;
     void stop() CV_OVERRIDE;
     bool isStarted() const CV_OVERRIDE;

modules/cudacodec/test/test_video.cpp

@@ -113,6 +113,10 @@ struct CheckParams : SetDevice
 {
 };
 
+struct Seek : SetDevice
+{
+};
+
 #if defined(HAVE_NVCUVID)
 //////////////////////////////////////////////////////
 // VideoReader
@@ -542,36 +546,22 @@ CUDA_TEST_P(CheckParams, Reader)
         ASSERT_TRUE(reader->get(cv::VideoCaptureProperties::CAP_PROP_OPEN_TIMEOUT_MSEC, msActual));
         ASSERT_EQ(msActual, msReference);
     }
-
-    {
-        std::vector<bool> exceptionsThrown = { false,true };
-        std::vector<int> capPropFormats = { -1,0 };
-        for (int i = 0; i < capPropFormats.size(); i++) {
-            bool exceptionThrown = false;
-            try {
-                cv::Ptr<cv::cudacodec::VideoReader> reader = cv::cudacodec::createVideoReader(inputFile, {
-                    cv::VideoCaptureProperties::CAP_PROP_FORMAT, capPropFormats.at(i) });
-            }
-            catch (cv::Exception &ex) {
-                if (ex.code == Error::StsUnsupportedFormat)
-                    exceptionThrown = true;
-            }
-            ASSERT_EQ(exceptionThrown, exceptionsThrown.at(i));
-        }
-    }
 }
 
 CUDA_TEST_P(CheckParams, CaptureProps)
 {
     std::string inputFile = std::string(cvtest::TS::ptr()->get_data_path()) + "../highgui/video/big_buck_bunny.mp4";
     cv::Ptr<cv::cudacodec::VideoReader> reader = cv::cudacodec::createVideoReader(inputFile);
-    double width, height, fps;
+    double width, height, fps, iFrame;
     ASSERT_TRUE(reader->get(cv::VideoCaptureProperties::CAP_PROP_FRAME_WIDTH, width));
     ASSERT_EQ(672, width);
     ASSERT_TRUE(reader->get(cv::VideoCaptureProperties::CAP_PROP_FRAME_HEIGHT, height));
     ASSERT_EQ(384, height);
     ASSERT_TRUE(reader->get(cv::VideoCaptureProperties::CAP_PROP_FPS, fps));
     ASSERT_EQ(24, fps);
+    ASSERT_TRUE(reader->grab());
+    ASSERT_TRUE(reader->get(cv::VideoCaptureProperties::CAP_PROP_POS_FRAMES, iFrame));
+    ASSERT_EQ(iFrame, 1.);
 }
 
 CUDA_TEST_P(CheckDecodeSurfaces, Reader)
@@ -619,6 +609,37 @@ CUDA_TEST_P(CheckInitParams, Reader)
     ASSERT_TRUE(reader->get(cv::cudacodec::VideoReaderProps::PROP_RAW_MODE, rawMode) && static_cast<bool>(rawMode) == params.rawMode);
 }
 
+CUDA_TEST_P(Seek, Reader)
+{
+#if defined(WIN32)
+    throw SkipTestException("Test disabled on Windows until the FFMpeg wrapper is updated to include PR24012.");
+#endif
+    std::string inputFile = std::string(cvtest::TS::ptr()->get_data_path()) + "../highgui/video/big_buck_bunny.mp4";
+    // seek to a non key frame
+    const int firstFrameIdx = 18;
+
+    GpuMat frameGs;
+    {
+        cv::Ptr<cv::cudacodec::VideoReader> readerGs = cv::cudacodec::createVideoReader(inputFile);
+        ASSERT_TRUE(readerGs->set(cudacodec::ColorFormat::GRAY));
+        for (int i = 0; i <= firstFrameIdx; i++)
+            ASSERT_TRUE(readerGs->nextFrame(frameGs));
+    }
+
+    cudacodec::VideoReaderInitParams params;
+    params.firstFrameIdx = firstFrameIdx;
+    cv::Ptr<cv::cudacodec::VideoReader> reader = cv::cudacodec::createVideoReader(inputFile, {}, params);
+    double iFrame = 0.;
+    ASSERT_TRUE(reader->get(cv::VideoCaptureProperties::CAP_PROP_POS_FRAMES, iFrame));
+    ASSERT_EQ(iFrame, static_cast<double>(firstFrameIdx));
+    ASSERT_TRUE(reader->set(cudacodec::ColorFormat::GRAY));
+    GpuMat frame;
+    ASSERT_TRUE(reader->nextFrame(frame));
+    ASSERT_EQ(cuda::norm(frameGs, frame, NORM_INF), 0.0);
+    ASSERT_TRUE(reader->get(cv::VideoCaptureProperties::CAP_PROP_POS_FRAMES, iFrame));
+    ASSERT_EQ(iFrame, static_cast<double>(firstFrameIdx+1));
+}
+
 #endif // HAVE_NVCUVID
 
 #if defined(HAVE_NVCUVID) && defined(HAVE_NVCUVENC)
@@ -958,5 +979,7 @@ INSTANTIATE_TEST_CASE_P(CUDA_Codec, CheckInitParams, testing::Combine(
     testing::Values("highgui/video/big_buck_bunny.mp4"),
     testing::Values(true,false), testing::Values(true,false), testing::Values(true,false)));
 
+INSTANTIATE_TEST_CASE_P(CUDA_Codec, Seek, ALL_DEVICES);
+
 #endif // HAVE_NVCUVID || HAVE_NVCUVENC
 }} // namespace