[FFmpeg-cvslog] dnn: add color conversion for analytic case

ffmpeg | branch: master | Guo, Yejun | Sun Feb 7 15:03:43 2021 +0800| [2da3a5c10f5249b21b8e37d1f354085178b0ffc3] | committer: Guo, Yejun dnn: add color conversion for analytic case Signed-off-by: Guo, Yejun >

[FFmpeg-cvslog] dnn_interface.h: add enum DNNColorOrder

ffmpeg | branch: master | Guo, Yejun | Mon Feb 8 09:53:39 2021 +0800| [0884063f88a9b726b296722cabe397b12337b16a] | committer: Guo, Yejun dnn_interface.h: add enum DNNColorOrder the data type and order together decide the color format, we could not use AVPixelFormat directly because not all

[FFmpeg-cvslog] dnn_backend_openvino.c: allow out_frame as NULL for analytic case

ffmpeg | branch: master | Ting Fu | Mon Feb 8 10:44:58 2021 +0800| [b0d75a8de924d90d0c9e8b75a485f35258920e02] | committer: Guo, Yejun dnn_backend_openvino.c: allow out_frame as NULL for analytic case >

[FFmpeg-cvslog] dnn: add function type for model

ffmpeg | branch: master | Guo, Yejun | Sun Feb 7 14:35:22 2021 +0800| [76fc6879e2c59929e7d81a72986bb4950053177e] | committer: Guo, Yejun dnn: add function type for model So the backend knows the usage of model is for frame processing, detect, classify, etc. Each function type has different

[FFmpeg-cvslog] dnn: extract common functions used by different filters

ffmpeg | branch: master | Guo, Yejun | Tue Jan 26 13:35:30 2021 +0800| [bdce6361004b3a24082e00b6c0126e2694a65b50] | committer: Guo, Yejun dnn: extract common functions used by different filters Signed-off-by: Guo, Yejun >

[FFmpeg-cvslog] dnn_backend_openvino.c: fix multi-thread issue for async execution

ffmpeg | branch: master | Guo, Yejun | Mon Feb 8 18:46:27 2021 +0800| [995c33a046bf058c79947755fc26c8fea5edb410] | committer: Guo, Yejun dnn_backend_openvino.c: fix multi-thread issue for async execution once we mark done for the task in function infer_completion_callback, the task is

[FFmpeg-cvslog] dnn_backend_openvino.c: fix mismatch between ffmpeg(NHWC) and openvino(NCHW)

ffmpeg | branch: master | Guo, Yejun | Mon Feb 8 09:22:23 2021 +0800| [51c105a62d931b6c27c8cad5b9aba3fd1de43668] | committer: Guo, Yejun dnn_backend_openvino.c: fix mismatch between ffmpeg(NHWC) and openvino(NCHW) Signed-off-by: Guo, Yejun >

[FFmpeg-cvslog] swscale/x86/swscale: fix compilation with old yasm

ffmpeg | branch: master | James Almer | Wed Feb 17 21:05:41 2021 -0300| [c2bf1dcace8d331e672b955f9cf5b59211749f00] | committer: James Almer swscale/x86/swscale: fix compilation with old yasm Where AVX2 may not be supported. Signed-off-by: James Almer >

[FFmpeg-cvslog] checkasm/sw_scale: use av_free() instead of free()

ffmpeg | branch: master | James Almer | Wed Feb 17 20:57:33 2021 -0300| [1371647fc381774d40b8beecf69604331262172d] | committer: James Almer checkasm/sw_scale: use av_free() instead of free() Fixes crashes on Win64 Signed-off-by: James Almer >

[FFmpeg-cvslog] avfilter/vf_pseudocolor: Add missing braces

ffmpeg | branch: master | Mark Thompson | Tue Feb 2 21:50:37 2021 +| [5a9aebac91e225440fd5f30897b88d6bdf265813] | committer: Mark Thompson avfilter/vf_pseudocolor: Add missing braces The array inside a structure needs two levels of braces. >

[FFmpeg-cvslog] swscale: move yuv2yuvX_sse3 to yasm, unrolls main loop

ffmpeg | branch: master | Alan Kelly | Thu Jan 14 15:47:03 2021 +0100| [554c2bc7086f49ef5a6a989ad6bc4bc11807eb6f] | committer: Paul B Mahol swscale: move yuv2yuvX_sse3 to yasm, unrolls main loop And other small optimizations for ~20% speedup. >

[FFmpeg-cvslog] x86/vf_gblur: fix reg name in UNIX64 prologue

ffmpeg | branch: master | James Almer | Wed Feb 17 15:51:28 2021 -0300| [1628409b18fb932b1ee0585151a82e67931cf43c] | committer: James Almer x86/vf_gblur: fix reg name in UNIX64 prologue Signed-off-by: James Almer >

[FFmpeg-cvslog] avfilter/vf_gblur: add missing arch check

ffmpeg | branch: master | James Almer | Wed Feb 17 15:45:40 2021 -0300| [3c77584be8fa6833b5cbf9b064ef1120a54fa61f] | committer: James Almer avfilter/vf_gblur: add missing arch check Removed by mistake in 2b4da1cb8c2984b37e5c912e103a1b8b734e7c1f where it should have been replaced instead.

[FFmpeg-cvslog] checkasm/vf_gblur: add a test for postscale_slice

ffmpeg | branch: master | James Almer | Wed Feb 17 12:05:12 2021 -0300| [bea7c513079a811512da378730366d80f8155f2d] | committer: James Almer checkasm/vf_gblur: add a test for postscale_slice Signed-off-by: James Almer >

[FFmpeg-cvslog] checkasm/vf_gblur: split off the horiz_slice test into its own function

ffmpeg | branch: master | James Almer | Wed Feb 17 12:04:23 2021 -0300| [2df3c2ed9b3ea1f2912bb663a5e942ba56385d05] | committer: James Almer checkasm/vf_gblur: split off the horiz_slice test into its own function Will come in handy for the following commit. Signed-off-by: James Almer >

[FFmpeg-cvslog] x86/vf_gblur: fix postscale_slice prologue

ffmpeg | branch: master | James Almer | Wed Feb 17 10:46:36 2021 -0300| [2b4da1cb8c2984b37e5c912e103a1b8b734e7c1f] | committer: James Almer x86/vf_gblur: fix postscale_slice prologue x86_32 ABI does not pass float arguments directly on xmm regs, and the Win64 ABI uses only the first four regs

[FFmpeg-cvslog] lavfi/drawtext: ignore final LF of textfile.

ffmpeg | branch: master | Nicolas George | Mon Nov 30 19:28:36 2020 +0100| [670051b52487cf71890909d1020a067169d743cc] | committer: Nicolas George lavfi/drawtext: ignore final LF of textfile. A standard text file ends with a final LF. Without this change, it is interpreted as an empty final