This commit fixes a wrong operation in the velocity conversion.
A reduced speed (because we are in a urban track) is now stored
into a uint8. Thus granularity is now half km/h.
Signed-off-by: Davide Sapienza <sapienza.dav@gmail.com>
This commit fixes a segmentation fault appeared in yolo3 network
updating. Now in Camera_t structure type there is a network pointer.
Signed-off-by: Davide Sapienza <sapienza.dav@gmail.com>
This commit changes the parameters reading. It introduces
getopt to read input parameters from command line and it
uses a configuration yaml file to read the input parameters
for the network, the map and the cameras.
This commit fixes a bug in message sending to the aggregator.
Signed-off-by: Davide Sapienza <sapienza.dav@gmail.com>
This commit pairs a camera with a thread. In this way, the
single process can to manage several camera. Every camera
thread create one video capture thread to read its input
stream. Only one camera thread can start the visualization.
This commit fixes the wrong data reading from file of the
read_projection_matrix function.
This commit fixes the wrong orientation mask accessing of
the addRoadUserfromTracker function. In the code there are
two sections to test.
This commit splits some operations into different threads.
Some threads compute the visualization preprocessing for the
live, detection, top view and disparity visualization.
Only one thread has the role to display the different views.
This commit moves the computation of the visualization
into the 'showImages' function (display thread). The main
thread workload and the time consuming for each frame are
reduced.
Please enter the commit message for your changes. Lines starting
This commit adds some box frame filters for the edge detection
(semantic segmentation) and the frame disparity operation,
both on the single frame box and on the whole image.
Micaela Verucchi
Francesco Gatti
luca
mbosi
Davide Sapienza
Davide Sapienza
Micaela Verucchi