How to Distinguish Slashed Zero From Eight (0->8) in OCR
I'm using ML Kit for Firebase for my Android app (ReCalc: Receipt Calculator) and it performs very well except in the case of slashed zero. In around half or more of the cases it recognizes slashed zero as eight.
One idea I have is to slice the rectangle containing the zero in regions and detect whether the regions just above and below the middle are dark or not. But actually...
I'm planning to train a model to classify zeroes and eights.
This is a lot of work thus I decided first to ask here for another solution/idea.
Here is an example:
Similar question: Tesseract OCR confuses slashed 0 as 8
Finally I've trained a model myself.
Its accuracy is pretty good (more than 98%). However I have concerns how well it generalize.
Here is the project: ZEC - Slashed Zero-Eight Classifier
I've created also an application showing how to use the model on Android: ZECA - Slashed Zero-Eight Classifier for Android
- Is there a way to set the data_min and the data_max in MinMaxScaler()?
- Do I need to use scaler even if my dataframe has fairly normalized data within a specific range
- What are the differences between fine tuning and few shot learning?
- Ordinal classification packages and algorithms
- ALS (Alternating Least Square) algorithm in multiple rankings for a user
- Keras, memoryerror - data = data.astype("float") / 255.0. Unable to allocate 309. MiB for an array with shape (13165, 32, 32, 3)
- Random Forest is overfitting
- How do cross_val_score and gridsearchCV work?
- what is difference between criterion and scoring in GridSearchCV
- Is a neural network a lazy or eager learning method?
- What is sharding in machine learning and how to do sharding in Tensorflow?
- Dataset for bank transaction
- How should a training dataset be distributed?
- How does Hydra `_partial_` interact with seeding
- StratifiedKFold vs KFold in scikit-learn
- Training a Keras model to identify leap years
- Ideas for Extracting Blade Tip Coordinates from masked Wind Turbine Image
- Macro VS Micro VS Weighted VS Samples F1 Score
- Doing PyWavelets calculation on GPU
- Training loss increases instead of decrease with epochs
- cannot access free variable 'fig' where it is not associated with a value in enclosing scope
- How to save a Dataset in multiple shards using `tf.data.Dataset.save`
- why explain logit as 'unscaled log probabililty' in sotfmax_cross_entropy_with_logits?
- What is the loss function used in Trainer from the Transformers library of Hugging Face?
- Sampling from image data
- Using features extracted using a pretrained CNN as new features for an CNN/NN
- InvalidArgumentError: No DNN in stream executor while training a TensorFlow RetinaNet model on Google Colab
- How does one set the pad token correctly (not to eos) during fine-tuning to avoid model not predicting EOS?
- How to create image of confusion matrix in Python
- Cross-validation with nb method