ML Model Accuracy/Results
Last updated
Last updated
Model details | Dataset | Accuracy achieved |
---|---|---|
Reasoning - As model is trained on the existing dataset, doesn't perform good on NIST dataset
Model details | Dataset | Accuracy achieved |
---|---|---|
Reasoning - After training the model on NIST misclassifications, improvement in accuracy
Model details | Dataset | Accuracy achieved |
---|---|---|
Reasoning - After training the model on NIST misclassifications as per previous checkpoint, improvement in accuracy
Model details | Dataset | Accuracy achieved |
---|---|---|
Reasoning - Training the model upon previous checkpoint and adding manually collected data, improvement in accuracy
Reasoning - As model is trained on the existing dataset, doesn't perform good on NIST dataset
Reasoning - After training the model on NIST misclassifications, improvement in accuracy
Reasoning: After training the model on NIST misclassifications and production dataset, improvement in accuracy
Reasoning: As averaging upon a large production dataset, the accuracy slightly dips as compared to iteration 3
Handwritten alphanumeric
Handwritten digits
Handwritten alphanumeric
Handwritten digits
Handwritten alphanumeric
Handwritten digits
Reasoning: Generally occurs if the digits are written in corners of the cell
Model details | Dataset | Accuracy achieved |
---|---|---|
Model details | Dataset | Accuracy achieved |
---|---|---|
Model details | Dataset | Accuracy achieved |
---|---|---|
Model details | Dataset | Accuracy achieved |
---|---|---|
Model trained and inference on
Existing dataset
99.80%
Model inference on
NIST dataset
62.70%
Average across all datasets
-
81.25%
Model trained on
Existing dataset + NIST misclassifications
83.30%
Model trained on
Existing dataset + NIST misclassifications
93.90%
Model trained on
Existing dataset + manually collected dataset from sheets (~1K samples)
93.90%
Model trained and inference on
Existing dataset
99.90%
Model inference on
NIST dataset
60.00%
Model inference on
Obtained production data (~50 samples)
97.70%
Average across all datasets
-
85.80%
Model trained on
Existing dataset + NIST misclasifications
96.40%
Model trained on
Existing dataset + NIST misclasifications + production dataset (~50 samples)
99.70%
Model trained on
Existing dataset + NIST misclasifications + manually collected dataset from sheets (~8.6k samples)
98.30%