-
Notifications
You must be signed in to change notification settings - Fork 20
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
chnsCompute ignores isInit #37
Comments
I believe That's probably what should happen here: acf/src/lib/acf/chnsCompute.cpp Lines 182 to 183 in 17b8da4
Ah. TODO :) This lib was created as part of an effort to get a few trained models running on a phone, so only some subset of the functionality has been tested. I'm not too surprised there are a number of use case gaps. Thanks for reporting. If you have a fix, feel free to send it. Otherwise, I can take a look later. We can add some GTests to cover some of this behavior as the issues come up. The tests are here: acf/src/lib/acf/ut/test-acf.cpp Lines 382 to 400 in 17b8da4
|
Don't have any proper fix as I'm handling the options on my own for now. There's a similar construct in |
To be true to the reference implementation I suppose we should aim for the following:
|
w/ |
* support chnsCompute + isInit: see #37 * support rgb-to-gray w/ correct channel count==1 : see #36 * `*.mat` serialization testing + download Inria/Caltech *.mat toolbox pedestrian models from github storage using cmake/hunter_private_data + add simple serialization tests in test-acf GTest code (more rigorous testing needed) + remove unused code blocks and some debug code in test-acf.cpp * fix a boundary case in sse gradHist() code * remove unused conditional cv::cvtColor() code blocks in rgbConvert.cpp * add virtual destructor to ObjectDetector.h base class * support acf shrink (via constructor input) in the GPUACF code * update gradhist glsl shader to highp precision * unify header include style * remove various comment blocks + format fixing * add acf special member isLuv to indicate special case input LUV processing (such as in gpgpu pipeline) * provide naive GLDetector : public acf::Detector naive detector to exercise shader code in acf-detect + use aglet lib for portable context + compute acf pyramids from input cv::Mat (in practice this would be separated for efficiency)
* unify header include style * opengl related acf fixes and … * support chnsCompute + isInit: see #37 * support rgb-to-gray w/ correct channel count==1 : see #36 * `*.mat` serialization testing + download Inria/Caltech *.mat toolbox pedestrian models from github storage using cmake/hunter_private_data + add simple serialization tests in test-acf GTest code (more rigorous testing needed) + remove unused code blocks and some debug code in test-acf.cpp * fix a boundary case in sse gradHist() code * remove unused conditional cv::cvtColor() code blocks in rgbConvert.cpp * add virtual destructor to ObjectDetector.h base class * support acf shrink (via constructor input) in the GPUACF code * update gradhist glsl shader to highp precision * unify header include style * remove various comment blocks + format fixing * add acf special member isLuv to indicate special case input LUV processing (such as in gpgpu pipeline) * provide naive GLDetector : public acf::Detector naive detector to exercise shader code in acf-detect + use aglet lib for portable context + compute acf pyramids from input cv::Mat (in practice this would be separated for efficiency) * clang-format * remove logging * remove output directory command line argument use working directory * test data download clarification * test data updates * tell user to update hunter if hunter_private_data is missing * now… assume test data is available in gauze_add_test * update drishti-upload submodule for newer hunter version so hunter_private_data is defined in ci builds * debug ci failure to read test image on some platforms * remove ci debugging * use PNG format test image Some of the platforms in drishti-upload are built without JPEG in OpenCV, so we use PNG format for now. The OpenCV CI configs can probably be updated and unified.
This should be consistent with the Matlab toolbox code now (after PR #40). I've added a couple tests to ensure that the default parameter retrieval is working and returning expected values. Closing ... |
https://github.com/elucideye/acf/blob/master/src/lib/acf/chnsCompute.cpp ignores the
isInit
-param so the options aren't returned. This seems to cause an issue withDetector::initializeOpts()
and later with functions likeDetector::computePyramid
which try to use the options.The text was updated successfully, but these errors were encountered: