Part Number: TDA2EXEVM
Hi,
First I runned semantic segmentation usecase completly correct with your test data files:inData and inHeader.
As suggestion ,I generate test data for semantic segmentation usecase as follow:
1、ffmpeg -i in3.mp4 -y -c libx264 -vf setsar=0:1 -refs:v 3 -r:v 15 -b 10M -threads 6 out2.h264 , the out2.h264 is inData for semantic segmentation usecase
2、PAT=size
sed -n "/$PAT/p" index.idx | cut -d' ' -f3 >> size_extracted.txt
3、delete all strings in size_exatracted.txt,then
cat size_extracted.txt | perl -n -e 'printf("%08x\n",$_)' | sed 's/\(..\)\(..\)\(..\)\(..\)/0x\4\n0x\3\n0x\2\n0x\1/' | xxd -r -p > Size_test.bin the Size_test.bin is inHeader for semantic segmentation usecase
the out2.h264 could play correct on PC,and when I use those tow files for semantic segmentation usecase,they can display and show results；
but the result show sequence is not correct,for examples,may be the frame 30 dispaly before the frame 29.
Is I misss some args for the command "ffmpeg" ?
Or what is the detail args for the "ffmpeg"?
Pls help me!It confuse me several days.
Best regards
