Other Parts Discussed in Thread: J721EXSOMXEVM,
Dear experts:
My camera is yuv format(nv12), based on OD demo code:app_tidl_od_cam,
what I would like to implement is (The flow is attached in an attached)Using Mosaic to merge 4 images the OD-20230118a.pdf:
Step (1): ReplaceMulti-scaler MSC by Mosaic MSC:
purpose: Using Mosaic to merge 4 input camera streams(1920x1080 each) into one stream with resolution 1024x512.
meaning that each camera is caling down to 512x256, then merge to a single 1024x512 frame.
Step (2) : Input the merge frame (1024x512) to pre-proc (C66-1) => Object Detection (C7x/MMA) => Post-proc(C66_2) <= same flow as TI demo code(app_tidl_od_cam)
Step (3) : Skip Mosaic MSC which follows the post-proc. and post-proc output directly to DSS.
My question are:
(1) Is step 1 doable? since Mosaic has functions of scaling down and merge.
(2) The input datatype of Pre-proc is ImgObj, which does not match the output format(vx_image) of Mosaic. I need to do formation transformation, How can I do?
(3) Regarding to format transformation from vx_image to ImgObj, what else I need to take care about???
Thanks.






_







