DoubangoTelecom/compv

View on GitHub

Insanely fast Open Source Computer Vision library for ARM and x86 devices (Up to #50 times faster than OpenCV)

Stars: 212
Forks: 47
Open beginner issues: 0
Indexed issues: 143

Dominant language: C++
License: GPL-3.0
Last GitHub push: Jul 30, 2026
Latest indexed: Sep 15, 2026
Contributing guide: No contributing guide
Code of conduct: No code of conduct
Beginner labels: No beginner labels indexed
PR merge metrics: No merged PRs in 30d

Open All

143 open issues indexed

Homography: split ransac in standalone implementation Open

DoubangoTelecom/compv#126 · 0 comments · 0 reactions · 0 assignees · Indexed Sep 15, 2026
Sobel could be done multiplication-free Open

DoubangoTelecom/compv#127 · 1 comment · 0 reactions · 0 assignees · Indexed Sep 15, 2026
No assembler functions for CompVCannyHysteresisRow_16mp_Intrin_SSE2 and CompVCannyHysteresisRow_16mp_Intrin_NEON Open

DoubangoTelecom/compv#128 · 0 comments · 0 reactions · 0 assignees · Indexed Sep 15, 2026
CompVCannyHysteresisRow_8mpw_Intrin_NEON is very slow Open

DoubangoTelecom/compv#129 · 0 comments · 0 reactions · 0 assignees · Indexed Sep 15, 2026
Canny: combine with mean instead of gaussian blur to remove noise Open

DoubangoTelecom/compv#130 · 0 comments · 0 reactions · 0 assignees · Indexed Sep 15, 2026
Houghstd, accum increment cache-unfriendly, this is why it's slow Open

DoubangoTelecom/compv#131 · 0 comments · 0 reactions · 0 assignees · Indexed Sep 15, 2026
OpenGL rendered images looks fade Open

DoubangoTelecom/compv#132 · 0 comments · 0 reactions · 0 assignees · Indexed Sep 15, 2026
Bundle (YUV -> RGB24) and (RGB24 -> HSV) and single MT instead of two MTs Open

DoubangoTelecom/compv#133 · 0 comments · 0 reactions · 0 assignees · Indexed Sep 15, 2026
SSE and AVX supports reciprocal (_mm_rcp_ps) use such instruction for scales (1/x) Open

DoubangoTelecom/compv#134 · 0 comments · 0 reactions · 0 assignees · Indexed Sep 15, 2026
Profile all AArch64 asm code with 'prfm pstl1keep' looks slower (MediaPad2) Open

DoubangoTelecom/compv#135 · 0 comments · 0 reactions · 0 assignees · Indexed Sep 15, 2026
Add ASM implementation for 'CompVHoughKhtPeaks_Section3_4_VotesCount_4mpd_Intrin_NEON' Open

DoubangoTelecom/compv#136 · 0 comments · 0 reactions · 0 assignees · Indexed Sep 15, 2026
Merge '__gauss_Eq15' with 'voting_Algorithm2_Gmin' and add SIMD implementation Open

DoubangoTelecom/compv#137 · 0 comments · 0 reactions · 0 assignees · Indexed Sep 15, 2026
Add SIMD implementation for 'CompVMathEigen<double>::find2x2' Open

DoubangoTelecom/compv#138 · 0 comments · 0 reactions · 0 assignees · Indexed Sep 15, 2026
Use MOVZX or MOVSX (al/ax/bx/bl/ah/bh...) to avoid "Partial register stall" issues Open

DoubangoTelecom/compv#141 · 1 comment · 0 reactions · 0 assignees · Indexed Sep 15, 2026
CompVImageScaleBilinear_Asm_XXX_YYY not faster than intrin implementations: no latency hiding technics used (rearrange instructions) Open

DoubangoTelecom/compv#143 · 1 comment · 0 reactions · 0 assignees · Indexed Sep 15, 2026
In arm assembler, replace "#define vecxxx vn" with "vecxxx .req vn" Open

DoubangoTelecom/compv#144 · 0 comments · 0 reactions · 0 assignees · Indexed Sep 15, 2026
Add AVX-512 implementation for histogram (gather/scatter) Open

DoubangoTelecom/compv#145 · 0 comments · 0 reactions · 0 assignees · Indexed Sep 15, 2026
posix_realloc_aligned -> relloc first then check if address changed Open

DoubangoTelecom/compv#146 · 0 comments · 0 reactions · 0 assignees · Indexed Sep 15, 2026
In all ARM intrin functions, use '__compv_builtin_assume_aligned' Open

DoubangoTelecom/compv#147 · 0 comments · 0 reactions · 0 assignees · Indexed Sep 15, 2026
compv_atomic_/(inc)/(dec) -> should lead to compilation error if neither visual studio nor gcc Open

DoubangoTelecom/compv#148 · 0 comments · 0 reactions · 0 assignees · Indexed Sep 15, 2026
Change pattern used in camera calibration: use more connected one Open

DoubangoTelecom/compv#149 · 0 comments · 0 reactions · 0 assignees · Indexed Sep 15, 2026
Add support for "ceres-solver" a Nonlinear Least Squares Minimizer Open

DoubangoTelecom/compv#150 · 0 comments · 0 reactions · 0 assignees · Indexed Sep 15, 2026
Camera calibration not working with 3264x2448 images (iPhone6) Open

DoubangoTelecom/compv#151 · 0 comments · 0 reactions · 0 assignees · Indexed Sep 15, 2026
2 remaps could be combined and done in one step Open

DoubangoTelecom/compv#152 · 0 comments · 0 reactions · 0 assignees · Indexed Sep 15, 2026
Add support for border type in convolution Open

DoubangoTelecom/compv#154 · 0 comments · 0 reactions · 0 assignees · Indexed Sep 15, 2026
Clean up gaussian kernel generation (see gabor for inspiration) Open

DoubangoTelecom/compv#155 · 0 comments · 0 reactions · 0 assignees · Indexed Sep 15, 2026
SSE41 and AVX2 no need for loading then unpack_zero for conversion -> direct memory conv exist Open

DoubangoTelecom/compv#156 · 0 comments · 0 reactions · 0 assignees · Indexed Sep 15, 2026
Replace all "vpmovmskb + test" with "vptest" Open

DoubangoTelecom/compv#157 · 0 comments · 0 reactions · 0 assignees · Indexed Sep 15, 2026
Use "CompVThreadDispatcher::dispatchDividingAcrossY" to avoid code verbosity Open

DoubangoTelecom/compv#159 · 0 comments · 0 reactions · 0 assignees · Indexed Sep 15, 2026
Call init in all newObj functions Open

DoubangoTelecom/compv#160 · 0 comments · 0 reactions · 0 assignees · Indexed Sep 15, 2026
Use "bsf" + "blsr" to loop through a mask from _mm_movemask_epi8 Open

DoubangoTelecom/compv#161 · 0 comments · 0 reactions · 0 assignees · Indexed Sep 15, 2026
ARM64 MediaPad2 -> MSER ST faster than MSER MT Open

DoubangoTelecom/compv#162 · 2 comments · 0 reactions · 0 assignees · Indexed Sep 15, 2026
Canny, should be "colStart = (imageWidthMinus1 & -mpw))" instead of "colStart = (imageWidthMinus1 & -(mpw - 1))" Open

DoubangoTelecom/compv#163 · 0 comments · 0 reactions · 0 assignees · Indexed Sep 15, 2026
Canny Hysteresis, use MatPtr as indexes, see UltBaseHysteresisProcessRow_8u_16mpb_Intrin_BSF_SSE2 Open

DoubangoTelecom/compv#164 · 0 comments · 0 reactions · 0 assignees · Indexed Sep 15, 2026
Add support for linear time morphological operator (for Toggle Mapping with large structuring element) Open

DoubangoTelecom/compv#165 · 0 comments · 0 reactions · 0 assignees · Indexed Sep 15, 2026
Change KNN item types from "int" to "size_t" Open

DoubangoTelecom/compv#166 · 0 comments · 0 reactions · 0 assignees · Indexed Sep 15, 2026
Bicubic transform very slooow (OpenCV faster) Open

DoubangoTelecom/compv#167 · 0 comments · 0 reactions · 0 assignees · Indexed Sep 15, 2026
Add support for LIGHTGBM Open

DoubangoTelecom/compv#168 · 0 comments · 0 reactions · 0 assignees · Indexed Sep 15, 2026
CompVMathExpExp_minpack1_64f64f_Asm_X64_AVX2 -> Intrin (VS2015) is faster than ASM Open

DoubangoTelecom/compv#169 · 0 comments · 0 reactions · 0 assignees · Indexed Sep 15, 2026
add support for metal (GPU, iOS, OSX) Open

DoubangoTelecom/compv#170 · 0 comments · 0 reactions · 0 assignees · Indexed Sep 15, 2026
Add support for POWER (ppc64el) Open

DoubangoTelecom/compv#171 · 0 comments · 0 reactions · 0 assignees · Indexed Sep 15, 2026
Add ASM code for CompVImageRemapBilinear_8u32f_Intrin Open

DoubangoTelecom/compv#172 · 0 comments · 0 reactions · 0 assignees · Indexed Sep 15, 2026
Drift in Bilinear rescale Open

DoubangoTelecom/compv#176 · 0 comments · 0 reactions · 0 assignees · Indexed Sep 15, 2026

DoubangoTelecom/compv

Get new issues in your inbox