s-trinh
commented
2 years ago Could you add a minimal code to reproduce the issue?
Closed erikstauber closed 2 years ago
s-trinh
commented
2 years ago Could you add a minimal code to reproduce the issue?
erikstauber
commented
2 years ago Yes, I'll prepare that..
erikstauber
commented
2 years ago <updated Feb 3rd 6am to reflect latest test data>
I've done a more comprehensive test using the tutorial code unmodified, and memory continues to get gobbled up by the program.
| ---------------------------------- | -----------memory usage at test time from Task Manager in MB-------- | Test- | -#Thrds- | -Detect Time(ms)- | -0min- | -10min- | -20min- | -30min- | -60min- | -120min- | -240min- | -900min- |
|---|---|---|---|---|---|---|---|---|---|---|---|---|
| 1--- | --1------ | --12-------------- | -67---- | -84----- | -101--- | -113---- | -108--- | ---------- | ---------- | --------- | ||
| 2--- | --4------ | --6--------------- | -72---- | -88----- | -104--- | -114---- | -147--- | -151----- | -282----- | -723---- |
Every now and then, memory usage temporarily goes down, but ultimately it continues to rise. This clearly looks like a bug.
fspindle
commented
2 years ago @erikstauber Can you check again if this issue is now fixed on your side using the master branch
I've seen a few older bugs related to memory leaks with the AprilTag detector; it still is leaking on my system. (Win 10 x64)
I'm running Visp3.4.0, and the method bool vpDetectorAprilTag::detect leaks about 20MB/min when detecting 5 tags at 60hz.