INITIALIZATION IS FAILING in both Keras and Darknet at the time of their classification. about labelreader HOT 22 CLOSED

It looks like you may need to recompile the corresponding binaries - they were compiled for my system, which will be different than yours.

On the other hand, I'd suggest using the docker image. It should be much easier to use as the dependencies are automatically installed.

from labelreader.

But when i make the docker image , and further use it
I have to clone the labelReader again and then it gives me another IMPORT ERROR for utlis
Please Guide on the same.

from labelreader.

You shouldn't be getting an import error if you've correctly downloaded the whole repo. Here are the steps I've just taken to confirm everything is working:

docker pull continuumio/miniconda3
docker run continuumio/miniconda3 /bin/bash
apt update
git clone https://github.com/ecthros/labelReader
cd labelReader
./install.sh
python labelReader.py -k -t

Can you confirm you are doing these same steps?

from labelreader.

Sure, I will try it and get back to you on the same.
Will run the commands on Ubuntu 16.04 , if any changes let me know.

from labelreader.

Tried configuring all of them , giving the same error

It should be python 2 or 3?

from labelreader.

It should be python3.

It looks like your config is failing with the error:
error: command 'gcc' failed with exit status 1.

Do you have gcc installed? Make sure to run install.sh, it should install gcc for you.

from labelreader.

Thanks , the error was with GCC only , but Keras is still failing in Initializing the Classifier.
Can you tell from where to pick the images , the specific folder or the location from where we can train our own data

from labelreader.

Keras is still failing in Initializing the Classifier.

from labelreader.

I've added some samples in the samples/ directory. Note that my model is trained on similar samples.
If you want to train your own model, I'd suggest looking at the README - it has steps on how to train for Darknet.
For the Keras classifier, I'm guessing it's missing a dependency. Please try to run install.sh to completion. If it's finishing, head over to the keras-yolo3 directory and try running python convert.py and python yolo.py. See if those have errors.

from labelreader.

Where is the samples directory ? and how should i access them ?
ok i will try keras one and get back to you

from labelreader.

Oops, forgot to push - it should be in the root of the repo now

from labelreader.

Thanks, I will clone it again and compile it and get back to you on the same.

from labelreader.

Didn't workout

from labelreader.

Looks like they've updated keras-yolo3. While I'm working on a permanent fix, do the following to revert to the earlier commit:

cd keras-yolo3
git checkout da7d756b0e47b979e701f0131ba7074ea138add8

Those steps should fix the error.
As a side note, you don't need both keras and darknet to work at the same time, only one.

from labelreader.

cd keras-yolo3 git checkout da7d756b0e47b979e701f0131ba7074ea138add8 ??

from labelreader.

That will check out the older version.

Alternatively, you don't have to run those commands - I've added the more permanent fix so you can just do a git pull again.

from labelreader.

On IT!

from labelreader.

Awesome, let me know when I can close this issue.

from labelreader.

If i have to train the darknet , i gotta make my own dataset right ?
Working with Keras as well , Thank YOU.

from labelreader.

Awesome. And yes, you'll need to train your own dataset, but you can follow the steps in the README to do so.

from labelreader.

Hey!

IT came with this issue now

from labelreader.

You can try changing the language for tesseract - it looks like it's using an incorrect language.

from labelreader.