Training problem about east HOT 7 CLOSED

hello @rmmal can you please share the code that you used to convert annotations?

from east.

I dont know, can you provide a training sample or you can check you annotations, see samples in training_samples/

from east.

it stuck alot in sleep(0.01)

I had a dataset that has x1,y1,x2,y2 and it's all about 1 class
so i converted it to
x1,y1,x2,y1,x2,y2,x1,y2 ,class1

for each image it may go through sleep(0.01) alot of times between reading the image in the generator function and the end of this function

what could be wrong ?

also i have a GPU that has 6GB ram would it be a problem ?

from east.

you may try to reduce the number of readers and batch size

from east.

@rmmal hello, how did you fix the problem about sleep(0.01) ?

from east.

@InesBeltaief if you have x1 , y1 , x2 , y2
then you can write them in such format
file_write.write(x1+","+y1+","+x2+","+y1+","+x2+","+y2+","+x1+","+y2+",text\n")

from east.

@yangxuehang there was a problem in my paths and dataset

from east.