profile
viewpoint

Ask questionsdouble free or corruption (!prev)

I get this error "double free or corruption (!prev)" when trying to start the training.

Do you know what is the issue? Might be related to memory usage, if so how can I change the batch size?

Best, Tommaso

dbolya/yolact

Answer questions TommasoBendinelli

I just checked the code. The issue comes from line 157 in augmentations.py. Also, I said that I commented out the line in my previous comment, but that is wrong. Sorry about that. The issue apparently comes from masks with third channel equal to 5 as mentioned in the other opencv issue...

I do not remember how I fix it, but what i can tell that now I am using width and height equal to max_size = 550 and with this setting it works

useful!

Related questions

KeyError on custom dataset hot 1
How do I run forward computing with C++? - yolact hot 1
RuntimeError: CUDA out of memory. Tried to allocate 20.00 MiB (GPU 0; 3.94 GiB total capacity; 3.36 GiB already allocated; 13.06 MiB free; 78.58 MiB cached) hot 1
What will happen if I change the backbone network into MobileNet-v2 with FPN? hot 1
Json file hot 1
??? a bug when i training hot 1
Training with custom data hot 1
TypeError: expected np.ndarray (got NoneType) hot 1
convert yolact to ONNX hot 1
RuntimeError: copy_if failed to synchronize: device-side assert triggered hot 1
How to run eval.py without cuda? hot 1
source:https://uonfu.com/
Github User Rank List