Displaying TI Deep Learning examples on x15 HDMI display

I’ve been working with the TI Deep Learning examples[1] on my x15 and have it working nicely by ssh’ing into it from a host computer and displaying the results on the host[2].

I now want to have the examples display their results on the x15’s hdmi display, but I’ve run into a problem. Here’s what I’m running:

x15$ cd tidl/tidl-api/examples/classification/
x15$ ./tidl_classification -g 1 -d 2 -e 2 -l ./imagenet.txt -s ./classlist.txt -i 0 -c ./stream_config_j11_v2.txt

If I run this logged in remotely via ssh -XC root@x15 it works fine. But it I run the x15 standalone with a monitor, mouse and keyboard and run the same commands I get the following error:

tidl_classification: /build/ti-llvm-3.6-ogtEpz/ti-llvm-3.6-3.6-git20170214/lib/Support/CommandLine.cpp:141: void llvm::cl::OptionCategory::registerCategory(): Assertion `std::count_if(RegisteredOptionCategories->begin(), RegisteredOptionCategories->end(), [this](const OptionCategory *Category) { return getName() == Category->getName(); }) == 0 && “Duplicate option categories”’ failed.

BUT, if I run and then quit gedit,

x15$ gedit

Error creating gpu
libGL error: failed to create dri screen
libGL error: failed to load driver: etnaviv

the above classification command works and the result is displayed on the hdmi monitor. So what is gedit doing that makes it work?


[1] http://downloads.ti.com/mctools/esd/docs/tidl-api/example.html#imagenet
[2] https://groups.google.com/forum/#!category-topic/beagleboard/efPr8XkNlj8
[3] https://www.google.com/url?q=https%3A%2F%2Felinux.org%2FEBC_Exercise_39_Setting_Up_tidl_on_X15&sa=D&sntz=1&usg=AFQjCNGYjnL7RZ1RsQqpyA9YhSHwPn0n5A