search

longphungtuan94 / ALPR_System

Automatic License Plate Recognition System for Vietnamese Plates
Apache License 2.0
92 stars 42 forks source link

ModuleNotFoundError: No module named 'utils.segmentation' #6

Closed prasetyoandi closed 4 years ago

prasetyoandi commented 4 years ago

i think everything is ok, module is also in the path. but this error still show up from from utils.segmentation import segment_characters_from_plate

longphungtuan94 commented 4 years ago

Hello. Which script did you run? Did you run from the ALPR_Sytem directory?

huygiangit96 commented 4 years ago

i think everything is ok, module is also in the path. but this error still show up from from utils.segmentation import segment_characters_from_plate

You can change the name of the "utils" folder like "_utils" then also change in code. I think it was conflict with "utils" in tensorflow.

prasetyoandi commented 4 years ago

i think everything is ok, module is also in the path. but this error still show up from from utils.segmentation import segment_characters_from_plate

You can change the name of the "utils" folder like "_utils" then also change in code. I think it was conflict with "utils" in tensorflow.

yea, thankyou i've missed it. but, mate. i have question, how to create custom char recognition model like binary_128_0.50_ver3.pb? @longphungtuan94 @huygiangit96

huygiangit96 commented 4 years ago

i think everything is ok, module is also in the path. but this error still show up from from utils.segmentation import segment_characters_from_plate

You can change the name of the "utils" folder like "_utils" then also change in code. I think it was conflict with "utils" in tensorflow.

yea, thankyou i've missed it. but, mate. i have question, how to create custom char recognition model like binary_128_0.50_ver3.pb? @longphungtuan94 @huygiangit96

You can follow this tutorial. It's quite simple https://tensorflow-object-detection-api-tutorial.readthedocs.io/en/latest/training.html#creating-label-map

prasetyoandi commented 4 years ago

i think everything is ok, module is also in the path. but this error still show up from from utils.segmentation import segment_characters_from_plate

You can change the name of the "utils" folder like "_utils" then also change in code. I think it was conflict with "utils" in tensorflow.

yea, thankyou i've missed it. but, mate. i have question, how to create custom char recognition model like binary_128_0.50_ver3.pb? @longphungtuan94 @huygiangit96

You can follow this tutorial. It's quite simple https://tensorflow-object-detection-api-tutorial.readthedocs.io/en/latest/training.html#creating-label-map

i think it's for object detection, not character recognition. maybe it's tensorflow attention_ocr, but it also can't read with my tf attention_ocr models.

huygiangit96 commented 4 years ago

I guessed that. I've never used before so I don't know exactly. Sorry

longphungtuan94 commented 4 years ago

i think everything is ok, module is also in the path. but this error still show up from from utils.segmentation import segment_characters_from_plate

You can change the name of the "utils" folder like "_utils" then also change in code. I think it was conflict with "utils" in tensorflow.

yea, thankyou i've missed it. but, mate. i have question, how to create custom char recognition model like binary_128_0.50_ver3.pb? @longphungtuan94 @huygiangit96

Hi @prasetyoandi . I created the character recognition model based on this tutorial: https://codelabs.developers.google.com/codelabs/tensorflow-for-poets/#0 It's a CNN architecture called MobileNet, which is quite appropriate for real time recognition.

prasetyoandi commented 4 years ago

i think everything is ok, module is also in the path. but this error still show up from from utils.segmentation import segment_characters_from_plate

You can change the name of the "utils" folder like "_utils" then also change in code. I think it was conflict with "utils" in tensorflow.

yea, thankyou i've missed it. but, mate. i have question, how to create custom char recognition model like binary_128_0.50_ver3.pb? @longphungtuan94 @huygiangit96

Hi @prasetyoandi . I created the character recognition model based on this tutorial: https://codelabs.developers.google.com/codelabs/tensorflow-for-poets/#0 It's a CNN architecture called MobileNet, which is quite appropriate for real time recognition.

is it means train every character(0-9 and A-Z)?, i don't get it

longphungtuan94 commented 4 years ago

Yes, I trained an Image Classifier for 36 classes (10 classes (0-9) and 26 classes (A-Z)). Otherwise, you could increase accuracy by training 2 independent models, the first one only for number classification (0-9) and the other for character classification (A-Z)

prasetyoandi commented 4 years ago

yes, thankyou very much for sharing.

charisma1919 commented 1 year ago

this tutorial not found can you suggest the other one? i search on google and cannot find the same article