Error during runing basic keras_ocr model in colab TypeError: annotate() missing 1 required positional argument: 's'

Question

I try to make a model to predict ocr text on documents. For this I use Keras_ocr. When I run code from documentation in google collab I get this error:

TypeError                                 Traceback (most recent call last)
<ipython-input-7-1df519f9081c> in <module>()
     19 fig, axs = plt.subplots(nrows=len(images), figsize=(20, 20))
     20 for ax, image, predictions in zip(axs, images, prediction_groups):
---> 21     keras_ocr.tools.drawAnnotations(image=image, predictions=predictions, ax=ax)

/usr/local/lib/python3.7/dist-packages/keras_ocr/tools.py in drawAnnotations(image, predictions, ax)
    182                 color="r",
    183                 fontsize=14,
--> 184                 horizontalalignment="right" if side == "left" else "left",
    185             )
    186     return ax

TypeError: annotate() missing 1 required positional argument: 's'

What is the problem ?

This is the link to documentation https://keras-ocr.readthedocs.io/en/latest/examples/using_pretrained_models.html

Answer 1

As per the recommendation from Github keras-ocr Issues 183. If you're getting annotate() missing 1 required positional argument: 's' , it means you are using old version of matplotlib. The input parameter changed from s to text in matplotlib v3.3.0.

To fix this, you have two options:

1. Upgrade matplotlib to >= 3.3.0, which supports the new text parameter.
2. Downgrade keras-ocr to <= v0.8.7, which uses the older s parameter.

In my case, following command in Google Colab worked for me

!pip install matplotlib==3.3.0