Commit Graph

157 Commits

Author SHA1 Message Date
Tadas Baltrusaitis
71dc98faa0 Some more removal of doubles. 2017-10-04 18:48:14 +01:00
Tadas Baltrusaitis
71174dc32b Moving more stuff over to floats (landmark locations and piece-wise affine warps) 2017-10-02 21:26:56 +01:00
Tadas Baltrusaitis
22d09a5193 Converting some more doubles to floats. 2017-10-01 13:22:03 +01:00
Tadas Baltrusaitis
f67cd346d2 Moving more of the PDM to floats 2017-10-01 12:56:14 +01:00
Tadas Baltrusaitis
30bedad382 Moving to floats in FaceAnalyser PDM. 2017-10-01 10:30:16 +01:00
Tadas Baltrusaitis
1197e5d1a1 Moving PDM internals to floats. 2017-10-01 10:09:32 +01:00
Tadas Baltrusaitis
156cf2e2dd Moving params_local to floats, preparing more things to move to floats. 2017-10-01 09:21:39 +01:00
Tadas Baltrusaitis
cedc430ef0 Simplifying the code a bit with moving to floats instead of doubles for likelihoods 2017-09-30 21:11:37 +01:00
Tadas Baltrusaitis
918651efdc Another small im2col speedup 2017-09-30 12:16:14 +01:00
Tadas Baltrusaitis
ea4a0feb3a Changing division to multiplication, a decent speedup. 2017-09-30 11:33:15 +01:00
Tadas Baltrusaitis
dae3816a06 Not spawning unnecessary threads with TBB. 2017-09-29 20:48:57 +01:00
Tadas Baltrusaitis
187e53f6e6 Quicker detection validation. 2017-09-24 20:34:35 +01:00
Tadas Baltrusaitis
15e18028f5 Optimizing CalcParams further. 2017-09-24 16:59:20 +01:00
Tadas Baltrusaitis
8c75c10dbf Speedup of CalcParams function. 2017-09-23 20:01:59 +01:00
Tadas Baltrusaitis
06c3aeb280 - AUs retrained using CE-CLM landmark detection
- bug fix with -gaze parameter
- bug fix with prefered location of bounding box
- re-initializing more often
2017-09-21 19:15:03 +01:00
Tadas Baltrusaitis
2928584e35 Sometimes direct iteration is slower for exp, so going to OpenCV implementation instead 2017-09-15 20:40:19 +01:00
Tadas Baltrusaitis
f7c09e0635 Bug fix with NU-RLMS optimization 2017-09-14 18:59:36 +01:00
Tadas Baltrusaitis
6c979c0bd3 Optimize NU-RLMS 2017-09-14 12:07:48 +01:00
Tadas Baltrusaitis
7d0e7e20a3 Optimizing model reading and some NU-RLMS computations using fast matrix multiplication. 2017-09-14 11:48:49 +01:00
Tadas Baltrusaitis
144c44bf61 Mirror id's bug fix with a constructor, quicker HOG model loading. 2017-09-14 08:10:03 +01:00
Tadas Baltrusaitis
62d4c4c8ce Speedup of CEN patch experts for frontal views. 2017-09-14 07:17:02 +01:00
Tadas Baltrusaitis
f5b364c54b - Choosing better CE-CLM models for C++ (more general and work well in a broader setting)
- Reducing the memory load of the model in C++ by half
2017-09-13 19:39:11 +01:00
Tadas Baltrusaitis
99a5fa160b Full run of evaluation suite using CE-CLM 2017-09-11 15:24:30 +01:00
Tadas Baltrusaitis
7e3a8570ca Copy constructor fix for face detection validator. 2017-09-03 08:47:56 +01:00
Tadas Baltrusaitis
25ea0d96b2 Copy constructor fix in MTCNN. 2017-09-02 20:15:05 +01:00
Tadas Baltrusaitis
ec691724ac Model fixes for CMake and BLAS fixes for Ubuntu. 2017-09-02 17:13:29 +01:00
Tadas Baltrusaitis
bac0083716 Fixes to cmake files for ubuntu, and fixes for gcc compiler. 2017-09-02 15:47:44 +01:00
Tadas Baltrusaitis
dadb255f43 Adding new files to cmake. 2017-09-02 09:28:16 +01:00
Tadas Baltrusaitis
c96aa6391b Dealing with unusual image formats (16 bit and 4 channel). 2017-09-02 09:01:24 +01:00
Tadas Baltrusaitis
f8e51bca28 Simplification and speedup of landmark validation. 2017-09-01 14:58:17 +01:00
Tadas Baltrusaitis
8e4ae1ace1 Another speedup with a fortran call. 2017-09-01 11:36:26 +01:00
Tadas Baltrusaitis
e0b8e775e7 Face detection speedup through faster convolution. 2017-09-01 11:19:45 +01:00
Tadas Baltrusaitis
3fcb3ce22b Head pose estimation with multiple hypothesis re-initialization (more accurate and can start from profile). 2017-08-31 20:13:21 +01:00
Tadas Baltrusaitis
c0d6306002 In case of MTCNN pick the biggest rather than most confident face. 2017-08-31 09:45:55 +01:00
Tadas Baltrusaitis
65172b8305 Some more speedup. 2017-08-30 20:01:25 +01:00
Tadas Baltrusaitis
53a1881ede Some cleanup. 2017-08-30 15:35:03 +01:00
Tadas Baltrusaitis
0ecf4e152c A bug fix 2017-08-30 15:25:26 +01:00
Tadas Baltrusaitis
b71b2f1cfc Another performance improvement by rearranging GEMM order. 2017-08-30 15:20:56 +01:00
Tadas Baltrusaitis
e1bf2c5a39 Another slight performance increase. 2017-08-29 21:43:27 +01:00
Tadas Baltrusaitis
cc3264e749 Another small speedup. 2017-08-29 11:39:18 +01:00
Tadas Baltrusaitis
4d4f0e47e1 Another speedup. 2017-08-29 08:08:20 +01:00
Tadas Baltrusaitis
d0d8a2d762 Another speedup, by optimizing the bias addition. 2017-08-29 07:55:39 +01:00
Tadas Baltrusaitis
8e2239ad81 Speeding up sigmoid. 2017-08-28 17:05:49 +01:00
Tadas Baltrusaitis
98e9d3e6f2 Faster way to call gemm through fortran. 2017-08-28 12:11:38 +01:00
Tadas Baltrusaitis
7e4109070b Big speedup of CNN using OpenBLAS, still lots of potential for speedups 2017-08-28 10:26:24 +01:00
Tadas Baltrusaitis
2b7620e56a Adding missing OpenBLAS binaries. 2017-08-28 08:32:26 +01:00
Tadas Baltrusaitis
c0b175c73a Adding missing OpenBLAS files 2017-08-28 08:27:27 +01:00
Tadas Baltrusaitis
6414efef72 - A better way to link to OpenBLAS (using pre-compiled binaries), this leads to faster computation on Windows
- Demo script fixes
2017-08-27 15:51:40 +01:00
Tadas Baltrusaitis
f62516789c Simplifying conversion to grayscale 2017-08-26 17:01:54 +01:00
Tadas Baltrusaitis
2ea630c834 Removing debug visualization. 2017-08-26 15:51:15 +01:00