Closed loharmain closed 3 years ago
@loharmain could you please share the training steps ? i want to train own recognition model using my dataset . training start with this repository but iteration/epoch not increase
could you please help me
try different setting such as changing width_ths
. See https://www.jaided.ai/easyocr/documentation/
So far I'm having very good results with 90% of the pictures that are similar (from same game and same coords). However, sometimes I experience digits being read and split into multiple parts. For example:
Becomes: ['56,383,248', '781', '150', '1,482,146', '3,890', '32,']
For some reason 32,781 are seperated?
As said before in 90% very similar cases there are no issues.
I'm using
.readtext(crop, detail = 0)