AnsweredAssumed Answered

How to use the apex_face_detection_cv.elf for large size picture?

Question asked by weikang zhang on Sep 4, 2018
Latest reply on Sep 6, 2018 by weikang zhang


    I used the demo apex_face_detection_cv.elf in vsdk and run it in my S32V234-EVB28899 board. I found that the picture used by the demo was in the path "/vsdk/data/common/input_faces.png". But I changed the default input_faces.png to a larger one, the error log was as below:

custom implementation finished

cascade classifier finished
OpenCV implementation finished
GenApuProgramParams() -> the input associated with port 'INPUT_IMAGE' is too wide (107 chunks) to fit into the CU array (64 CUs)
ACF_Process_APU::Start_Setup() -> program parameter generation failed
ACF_Process_APU::Wait() -> process was never started (i.e. nothing to wait for)
APEX implementation finished
custom host implementation: 4.328749s.
cascade classifier implementation: 0.011766s.
APEX implementation: 0.000118s.
OpenCV implementation: 0.035100s.
Framebuffer mapped at 0x7fa132c000.


 Please tell me how to change the source code for using a larger picture such as the "input_faces_multiscale.png" in the path "/vsdk/data/common/.