Add decoder APIs and unit tests in tile-coding experiment
In the tile-coding experiment,
1. In tile decoder, added 2 set control APIs:
VP10_SET_DECODE_TILE_ROW and VP10_SET_DECODE_TILE_COL. It allowed
users to set the range of decoding at frame level.
2. Added a unit test while tile-coding experiment is on. It tested
both tile encoder and decoder to make sure the encoded frame
can be decoded as a whole frame or as independent tiles.
Change-Id: I73fd0632b685047cb9376008127cde72efa3fb2b
diff --git a/test/decode_test_driver.cc b/test/decode_test_driver.cc
index ad861c3..7fb3e37 100644
--- a/test/decode_test_driver.cc
+++ b/test/decode_test_driver.cc
@@ -18,6 +18,7 @@
namespace libvpx_test {
const char kVP8Name[] = "WebM Project VP8";
+const char kVP10Name[] = "WebM Project VP10";
vpx_codec_err_t Decoder::PeekStream(const uint8_t *cxdata, size_t size,
vpx_codec_stream_info_t *stream_info) {
@@ -46,6 +47,11 @@
return strncmp(kVP8Name, codec_name, sizeof(kVP8Name) - 1) == 0;
}
+bool Decoder::IsVP10() const {
+ const char *codec_name = GetDecoderName();
+ return strncmp(kVP10Name, codec_name, sizeof(kVP10Name) - 1) == 0;
+}
+
void DecoderTest::HandlePeekResult(Decoder *const decoder,
CompressedVideoSource *video,
const vpx_codec_err_t res_peek) {
diff --git a/test/decode_test_driver.h b/test/decode_test_driver.h
index f566c53..1492c5a 100644
--- a/test/decode_test_driver.h
+++ b/test/decode_test_driver.h
@@ -107,6 +107,8 @@
bool IsVP8() const;
+ bool IsVP10() const;
+
vpx_codec_ctx_t * GetDecoder() {
return &decoder_;
}
diff --git a/test/encode_test_driver.cc b/test/encode_test_driver.cc
index e24c9bf..f3c3243 100644
--- a/test/encode_test_driver.cc
+++ b/test/encode_test_driver.cc
@@ -56,7 +56,7 @@
#endif // !CONFIG_EXT_TILE
} else
#endif
- {
+ if (CodecInterface() == &vpx_codec_vp8_cx_algo) {
#if CONFIG_VP8_ENCODER
ASSERT_EQ(&vpx_codec_vp8_cx_algo, CodecInterface())
<< "Unknown Codec Interface";
@@ -261,12 +261,6 @@
void EncoderTest::RunLoop(VideoSource *video) {
vpx_codec_dec_cfg_t dec_cfg = vpx_codec_dec_cfg_t();
-#if CONFIG_EXT_TILE
- // Decode all tiles.
- dec_cfg.tile_col = -1;
- dec_cfg.tile_row = -1;
-#endif // CONFIG_EXT_TILE
-
stats_.Reset();
ASSERT_TRUE(passes_ == 1 || passes_ == 2);
@@ -295,6 +289,15 @@
if (init_flags_ & VPX_CODEC_USE_OUTPUT_PARTITION)
dec_init_flags |= VPX_CODEC_USE_INPUT_FRAGMENTS;
Decoder* const decoder = codec_->CreateDecoder(dec_cfg, dec_init_flags, 0);
+#if CONFIG_VP10 && CONFIG_EXT_TILE
+ if (decoder->IsVP10()) {
+ // Set dec_cfg.tile_row = -1 and dec_cfg.tile_col = -1 so that the whole
+ // frame is decoded.
+ decoder->Control(VP10_SET_DECODE_TILE_ROW, -1);
+ decoder->Control(VP10_SET_DECODE_TILE_COL, -1);
+ }
+#endif
+
bool again;
for (again = true; again; video->Next()) {
again = (video->img() != NULL);
diff --git a/test/svc_test.cc b/test/svc_test.cc
index e573e10..1ad17be 100644
--- a/test/svc_test.cc
+++ b/test/svc_test.cc
@@ -61,12 +61,14 @@
codec_enc_.kf_max_dist = 100;
vpx_codec_dec_cfg_t dec_cfg = vpx_codec_dec_cfg_t();
-#if CONFIG_EXT_TILE
- dec_cfg.tile_col = -1;
- dec_cfg.tile_row = -1;
-#endif // CONFIG_EXT_TILE
VP9CodecFactory codec_factory;
decoder_ = codec_factory.CreateDecoder(dec_cfg, 0);
+#if CONFIG_VP10 && CONFIG_EXT_TILE
+ if (decoder_->IsVP10()) {
+ decoder_->Control(VP10_SET_DECODE_TILE_ROW, -1);
+ decoder_->Control(VP10_SET_DECODE_TILE_COL, -1);
+ }
+#endif
tile_columns_ = 0;
tile_rows_ = 0;
diff --git a/test/test.mk b/test/test.mk
index 59f054c..7412b27 100644
--- a/test/test.mk
+++ b/test/test.mk
@@ -170,6 +170,7 @@
LIBVPX_TEST_SRCS-$(CONFIG_VP10_ENCODER) += vp10_fht8x8_test.cc
LIBVPX_TEST_SRCS-$(CONFIG_VP10_ENCODER) += vp10_fht16x16_test.cc
LIBVPX_TEST_SRCS-$(CONFIG_ANS) += vp10_ans_test.cc
+LIBVPX_TEST_SRCS-$(CONFIG_EXT_TILE) += vp10_ext_tile_test.cc
LIBVPX_TEST_SRCS-$(CONFIG_VP10_ENCODER) += sum_squares_test.cc
LIBVPX_TEST_SRCS-$(CONFIG_VP10_ENCODER) += subtract_test.cc
diff --git a/test/tile_independence_test.cc b/test/tile_independence_test.cc
index 9a049bf..dc31d06 100644
--- a/test/tile_independence_test.cc
+++ b/test/tile_independence_test.cc
@@ -35,13 +35,18 @@
cfg.w = 704;
cfg.h = 144;
cfg.threads = 1;
-#if CONFIG_EXT_TILE
- cfg.tile_col = -1;
- cfg.tile_row = -1;
-#endif // CONFIG_EXT_TILE
fw_dec_ = codec_->CreateDecoder(cfg, 0);
inv_dec_ = codec_->CreateDecoder(cfg, 0);
inv_dec_->Control(VP9_INVERT_TILE_DECODE_ORDER, 1);
+
+#if CONFIG_VP10 && CONFIG_EXT_TILE
+ if (fw_dec_->IsVP10() && inv_dec_->IsVP10()) {
+ fw_dec_->Control(VP10_SET_DECODE_TILE_ROW, -1);
+ fw_dec_->Control(VP10_SET_DECODE_TILE_COL, -1);
+ inv_dec_->Control(VP10_SET_DECODE_TILE_ROW, -1);
+ inv_dec_->Control(VP10_SET_DECODE_TILE_COL, -1);
+ }
+#endif
}
virtual ~TileIndependenceTest() {
diff --git a/test/vp10_ext_tile_test.cc b/test/vp10_ext_tile_test.cc
new file mode 100644
index 0000000..7c9e960
--- /dev/null
+++ b/test/vp10_ext_tile_test.cc
@@ -0,0 +1,201 @@
+/*
+ * Copyright (c) 2016 The WebM project authors. All Rights Reserved.
+ *
+ * Use of this source code is governed by a BSD-style license
+ * that can be found in the LICENSE file in the root of the source
+ * tree. An additional intellectual property rights grant can be found
+ * in the file PATENTS. All contributing project authors may
+ * be found in the AUTHORS file in the root of the source tree.
+ */
+
+#include <assert.h>
+#include <string>
+#include <vector>
+#include "third_party/googletest/src/include/gtest/gtest.h"
+#include "test/codec_factory.h"
+#include "test/encode_test_driver.h"
+#include "test/i420_video_source.h"
+#include "test/md5_helper.h"
+#include "test/util.h"
+
+namespace {
+// The number of frames to be encoded/decoded
+const int kLimit = 8;
+// Skip 1 frame to check the frame decoding independency.
+const int kSkip = 5;
+const int kTileSize = 1;
+const int kTIleSizeInPixels = (kTileSize << 6);
+// Fake width and height so that they can be multiples of the tile size.
+const int kImgWidth = 704;
+const int kImgHeight = 576;
+
+class VP10ExtTileTest
+ : public ::libvpx_test::EncoderTest,
+ public ::libvpx_test::CodecTestWith2Params<libvpx_test::TestMode, int> {
+ protected:
+ VP10ExtTileTest()
+ : EncoderTest(GET_PARAM(0)),
+ encoding_mode_(GET_PARAM(1)),
+ set_cpu_used_(GET_PARAM(2)) {
+ init_flags_ = VPX_CODEC_USE_PSNR;
+ vpx_codec_dec_cfg_t cfg = vpx_codec_dec_cfg_t();
+ cfg.w = kImgWidth;
+ cfg.h = kImgHeight;
+
+ decoder_ = codec_->CreateDecoder(cfg, 0);
+ decoder_->Control(VP10_SET_DECODE_TILE_ROW, -1);
+ decoder_->Control(VP10_SET_DECODE_TILE_COL, -1);
+
+ // Allocate buffer to store tile image.
+ vpx_img_alloc(&tile_img_, VPX_IMG_FMT_I420, kImgWidth, kImgHeight, 32);
+
+ md5_.clear();
+ tile_md5_.clear();
+ }
+
+ virtual ~VP10ExtTileTest() {
+ vpx_img_free(&tile_img_);
+ delete decoder_;
+ }
+
+ virtual void SetUp() {
+ InitializeConfig();
+ SetMode(encoding_mode_);
+
+ cfg_.g_lag_in_frames = 0;
+ cfg_.rc_end_usage = VPX_VBR;
+ cfg_.g_error_resilient = 1;
+
+ cfg_.rc_max_quantizer = 56;
+ cfg_.rc_min_quantizer = 0;
+ }
+
+ virtual void PreEncodeFrameHook(::libvpx_test::VideoSource * video,
+ ::libvpx_test::Encoder *encoder) {
+ if (video->frame() == 0) {
+ // Encode setting
+ encoder->Control(VP8E_SET_CPUUSED, set_cpu_used_);
+ encoder->Control(VP8E_SET_ENABLEAUTOALTREF, 0);
+ encoder->Control(VP9E_SET_FRAME_PARALLEL_DECODING, 1);
+
+ // The tile size is 64x64.
+ encoder->Control(VP9E_SET_TILE_COLUMNS, kTileSize);
+ encoder->Control(VP9E_SET_TILE_ROWS, kTileSize);
+#if CONFIG_EXT_PARTITION
+ // Always use 64x64 max partition.
+ encoder->Control(VP10E_SET_SUPERBLOCK_SIZE, VPX_SUPERBLOCK_SIZE_64X64);
+#endif
+ }
+
+ if (video->frame() == 1) {
+ frame_flags_ = VP8_EFLAG_NO_UPD_LAST | VP8_EFLAG_NO_UPD_GF |
+ VP8_EFLAG_NO_UPD_ARF;
+ }
+ }
+
+ virtual void DecompressedFrameHook(const vpx_image_t &img,
+ vpx_codec_pts_t pts) {
+ // Skip 1 already decoded frame to be consistent with the decoder in this
+ // test.
+ if (pts == (vpx_codec_pts_t)kSkip)
+ return;
+
+ // Calculate MD5 as the reference.
+ ::libvpx_test::MD5 md5_res;
+ md5_res.Add(&img);
+ md5_.push_back(md5_res.Get());
+ }
+
+ virtual void FramePktHook(const vpx_codec_cx_pkt_t *pkt) {
+ // Skip decoding 1 frame.
+ if (pkt->data.frame.pts == (vpx_codec_pts_t)kSkip)
+ return;
+
+ bool IsLastFrame = (pkt->data.frame.pts == (vpx_codec_pts_t)(kLimit - 1));
+
+ // Decode the first (kLimit - 1) frames as whole frame, and decode the last
+ // frame in single tiles.
+ for (int r = 0; r < kImgHeight / kTIleSizeInPixels; ++r) {
+ for (int c = 0; c < kImgWidth / kTIleSizeInPixels; ++c) {
+ if (!IsLastFrame) {
+ decoder_->Control(VP10_SET_DECODE_TILE_ROW, -1);
+ decoder_->Control(VP10_SET_DECODE_TILE_COL, -1);
+ } else {
+ decoder_->Control(VP10_SET_DECODE_TILE_ROW, r);
+ decoder_->Control(VP10_SET_DECODE_TILE_COL, c);
+ }
+
+ const vpx_codec_err_t res = decoder_->DecodeFrame(
+ reinterpret_cast<uint8_t*>(pkt->data.frame.buf),
+ pkt->data.frame.sz);
+ if (res != VPX_CODEC_OK) {
+ abort_ = true;
+ ASSERT_EQ(VPX_CODEC_OK, res);
+ }
+ const vpx_image_t *img = decoder_->GetDxData().Next();
+
+ if (!IsLastFrame) {
+ if (img) {
+ ::libvpx_test::MD5 md5_res;
+ md5_res.Add(img);
+ tile_md5_.push_back(md5_res.Get());
+ }
+ break;
+ }
+
+ const int kMaxMBPlane = 3;
+ for (int plane = 0; plane < kMaxMBPlane; ++plane) {
+ const int shift = (plane == 0) ? 0 : 1;
+ int tile_height = kTIleSizeInPixels >> shift;
+ int tile_width = kTIleSizeInPixels >> shift;
+
+ for (int tr = 0; tr < tile_height; ++tr) {
+ memcpy(tile_img_.planes[plane] +
+ tile_img_.stride[plane] * (r * tile_height + tr) +
+ c * tile_width,
+ img->planes[plane] + img->stride[plane] * tr, tile_width);
+ }
+ }
+ }
+
+ if (!IsLastFrame)
+ break;
+ }
+
+ if (IsLastFrame && &tile_img_) {
+ ::libvpx_test::MD5 md5_res;
+ md5_res.Add(&tile_img_);
+ tile_md5_.push_back(md5_res.Get());
+ }
+ }
+
+ ::libvpx_test::TestMode encoding_mode_;
+ int set_cpu_used_;
+ ::libvpx_test::Decoder *decoder_;
+ vpx_image_t tile_img_;
+ std::vector<std::string> md5_;
+ std::vector<std::string> tile_md5_;
+};
+
+TEST_P(VP10ExtTileTest, DecoderResultTest) {
+ ::libvpx_test::I420VideoSource video("hantro_collage_w352h288.yuv",
+ kImgWidth, kImgHeight, 30, 1, 0, kLimit);
+ cfg_.rc_target_bitrate = 500;
+ cfg_.g_error_resilient = VPX_ERROR_RESILIENT_DEFAULT;
+ cfg_.g_lag_in_frames = 0;
+ cfg_.g_threads = 1;
+
+ // Tile encoding
+ init_flags_ = VPX_CODEC_USE_PSNR;
+ ASSERT_NO_FATAL_FAILURE(RunLoop(&video));
+
+ // Compare to check if two vectors are equal.
+ ASSERT_EQ(md5_, tile_md5_);
+}
+
+VP10_INSTANTIATE_TEST_CASE(
+ // Now only test 2-pass mode.
+ VP10ExtTileTest,
+ ::testing::Values(::libvpx_test::kTwoPassGood),
+ ::testing::Range(0, 4));
+} // namespace
diff --git a/test/vp9_ethread_test.cc b/test/vp9_ethread_test.cc
index d6b6951..35a6619 100644
--- a/test/vp9_ethread_test.cc
+++ b/test/vp9_ethread_test.cc
@@ -33,11 +33,13 @@
vpx_codec_dec_cfg_t cfg = vpx_codec_dec_cfg_t();
cfg.w = 1280;
cfg.h = 720;
-#if CONFIG_EXT_TILE
- cfg.tile_col = -1;
- cfg.tile_row = -1;
-#endif // CONFIG_EXT_TILE
decoder_ = codec_->CreateDecoder(cfg, 0);
+#if CONFIG_VP10 && CONFIG_EXT_TILE
+ if (decoder_->IsVP10()) {
+ decoder_->Control(VP10_SET_DECODE_TILE_ROW, -1);
+ decoder_->Control(VP10_SET_DECODE_TILE_COL, -1);
+ }
+#endif
size_enc_.clear();
md5_dec_.clear();
diff --git a/vp10/vp10_dx_iface.c b/vp10/vp10_dx_iface.c
index cf6ab56..8dc49f2 100644
--- a/vp10/vp10_dx_iface.c
+++ b/vp10/vp10_dx_iface.c
@@ -58,6 +58,8 @@
int last_show_frame; // Index of last output frame.
int byte_alignment;
int skip_loop_filter;
+ int decode_tile_row;
+ int decode_tile_col;
// Frame parallel related.
int frame_parallel_decode; // frame-based threading.
@@ -501,8 +503,8 @@
frame_worker_data->pbi->decrypt_state = ctx->decrypt_state;
#if CONFIG_EXT_TILE
- frame_worker_data->pbi->dec_tile_row = ctx->cfg.tile_row;
- frame_worker_data->pbi->dec_tile_col = ctx->cfg.tile_col;
+ frame_worker_data->pbi->dec_tile_row = ctx->decode_tile_row;
+ frame_worker_data->pbi->dec_tile_col = ctx->decode_tile_col;
#endif // CONFIG_EXT_TILE
worker->had_error = 0;
@@ -1118,6 +1120,18 @@
return VPX_CODEC_OK;
}
+static vpx_codec_err_t ctrl_set_decode_tile_row(vpx_codec_alg_priv_t *ctx,
+ va_list args) {
+ ctx->decode_tile_row = va_arg(args, int);
+ return VPX_CODEC_OK;
+}
+
+static vpx_codec_err_t ctrl_set_decode_tile_col(vpx_codec_alg_priv_t *ctx,
+ va_list args) {
+ ctx->decode_tile_col = va_arg(args, int);
+ return VPX_CODEC_OK;
+}
+
static vpx_codec_ctrl_fn_map_t decoder_ctrl_maps[] = {
{VP8_COPY_REFERENCE, ctrl_copy_reference},
@@ -1132,6 +1146,8 @@
{VPXD_SET_DECRYPTOR, ctrl_set_decryptor},
{VP9_SET_BYTE_ALIGNMENT, ctrl_set_byte_alignment},
{VP9_SET_SKIP_LOOP_FILTER, ctrl_set_skip_loop_filter},
+ {VP10_SET_DECODE_TILE_ROW, ctrl_set_decode_tile_row},
+ {VP10_SET_DECODE_TILE_COL, ctrl_set_decode_tile_col},
// Getters
{VP8D_GET_LAST_REF_UPDATES, ctrl_get_last_ref_updates},
diff --git a/vpx/vp8dx.h b/vpx/vp8dx.h
index 1f02fd5..347521e 100644
--- a/vpx/vp8dx.h
+++ b/vpx/vp8dx.h
@@ -121,7 +121,16 @@
*/
VP9_SET_SKIP_LOOP_FILTER,
- VP8_DECODER_CTRL_ID_MAX
+ VP8_DECODER_CTRL_ID_MAX,
+
+ /** control function to set the range of tile decoding. A value that is
+ * greater and equal to zero indicates only the specific row/column is
+ * decoded. A value that is -1 indicates the whole row/column is decoded.
+ * A special case is both values are -1 that means the whole frame is
+ * decoded.
+ */
+ VP10_SET_DECODE_TILE_ROW,
+ VP10_SET_DECODE_TILE_COL
};
/** Decrypt n bytes of data from input -> output, using the decrypt_state
@@ -174,7 +183,10 @@
#define VPX_CTRL_VP9D_GET_FRAME_SIZE
VPX_CTRL_USE_TYPE(VP9_INVERT_TILE_DECODE_ORDER, int)
#define VPX_CTRL_VP9_INVERT_TILE_DECODE_ORDER
-
+VPX_CTRL_USE_TYPE(VP10_SET_DECODE_TILE_ROW, int)
+#define VPX_CTRL_VP10_SET_DECODE_TILE_ROW
+VPX_CTRL_USE_TYPE(VP10_SET_DECODE_TILE_COL, int)
+#define VPX_CTRL_VP10_SET_DECODE_TILE_COL
/*!\endcond */
/*! @} - end defgroup vp8_decoder */
diff --git a/vpx/vpx_decoder.h b/vpx/vpx_decoder.h
index d4ba986..62fd919 100644
--- a/vpx/vpx_decoder.h
+++ b/vpx/vpx_decoder.h
@@ -108,10 +108,6 @@
unsigned int threads; /**< Maximum number of threads to use, default 1 */
unsigned int w; /**< Width */
unsigned int h; /**< Height */
- int tile_row; /**< The index of row tile to be decoded.
- Value -1 means to decode all row tiles. */
- int tile_col; /**< The index of column tile to be decoded.
- Value -1 means to decode all column tiles */
} vpx_codec_dec_cfg_t; /**< alias for struct vpx_codec_dec_cfg */
diff --git a/vpxdec.c b/vpxdec.c
index 13b020b..235d17a 100644
--- a/vpxdec.c
+++ b/vpxdec.c
@@ -818,11 +818,6 @@
if (!interface)
interface = get_vpx_decoder_by_index(0);
-#if CONFIG_EXT_TILE
- cfg.tile_row = tile_row;
- cfg.tile_col = tile_col;
-#endif // CONFIG_EXT_TILE
-
dec_flags = (postproc ? VPX_CODEC_USE_POSTPROC : 0) |
(ec_enabled ? VPX_CODEC_USE_ERROR_CONCEALMENT : 0) |
(frame_parallel ? VPX_CODEC_USE_FRAME_THREADING : 0);
@@ -877,6 +872,21 @@
}
#endif
+#if CONFIG_VP10_DECODER && CONFIG_EXT_TILE
+ if (strncmp(decoder.name, "WebM Project VP10", 17) == 0) {
+ if (vpx_codec_control(&decoder, VP10_SET_DECODE_TILE_ROW, tile_row)) {
+ fprintf(stderr, "Failed to set decode_tile_row: %s\n",
+ vpx_codec_error(&decoder));
+ return EXIT_FAILURE;
+ }
+
+ if (vpx_codec_control(&decoder, VP10_SET_DECODE_TILE_COL, tile_col)) {
+ fprintf(stderr, "Failed to set decode_tile_col: %s\n",
+ vpx_codec_error(&decoder));
+ return EXIT_FAILURE;
+ }
+ }
+#endif
if (arg_skip)
fprintf(stderr, "Skipping first %d frames.\n", arg_skip);
diff --git a/vpxenc.c b/vpxenc.c
index 7fb28cd..9a52d54 100644
--- a/vpxenc.c
+++ b/vpxenc.c
@@ -1580,8 +1580,18 @@
#if CONFIG_DECODERS
if (global->test_decode != TEST_DECODE_OFF) {
const VpxInterface *decoder = get_vpx_decoder_by_name(global->codec->name);
- vpx_codec_dec_cfg_t cfg = { 0, 0, 0, -1, -1 };
+ vpx_codec_dec_cfg_t cfg = { 0, 0, 0};
vpx_codec_dec_init(&stream->decoder, decoder->codec_interface(), &cfg, 0);
+
+#if CONFIG_VP10_DECODER && CONFIG_EXT_TILE
+ if (strcmp(global->codec->name, "vp10") == 0) {
+ vpx_codec_control(&stream->decoder, VP10_SET_DECODE_TILE_ROW, -1);
+ ctx_exit_on_error(&stream->decoder, "Failed to set decode_tile_row");
+
+ vpx_codec_control(&stream->decoder, VP10_SET_DECODE_TILE_COL, -1);
+ ctx_exit_on_error(&stream->decoder, "Failed to set decode_tile_col");
+ }
+#endif
}
#endif
}
