Layoutlmv3 Finetuned Wildreceipt

🚀 layoutlmv3-finetuned-wildreceipt

This model is a fine - tuned version of [microsoft/layoutlmv3 - base](https://huggingface.co/microsoft/layoutlmv3 - base) on the wild_receipt dataset. It is used for key information extraction from receipt images, achieving high precision, recall, F1 score, and accuracy on the evaluation set.

🚀 Quick Start

This section is not provided in the original README, so it is skipped.

✨ Features

This model is fine - tuned on the wild_receipt dataset, which is specifically designed for key information extraction from receipts. It has achieved excellent performance on multiple evaluation metrics such as precision, recall, F1 score, and accuracy.

📚 Documentation

Model Performance

This model achieves the following results on the evaluation set:

• Loss: 0.3108
• Precision: 0.8772
• Recall: 0.8799
• F1: 0.8785
• Accuracy: 0.9249

Training and Evaluation Data

The WildReceipt dataset consists of 1740 receipt images, and contains 25 key information categories, and a total of about 69000 text boxes. 1268 and 472 images are used for training and testing respectively to train the LayoutLMv3 model for Key Information Extraction.

Training procedure

The training code: https://github.com/Theivaprakasham/layoutlmv3/blob/main/training_codes/LayoutLMv3_training_WildReceipts_dataset.ipynb

Training hyperparameters

The following hyperparameters were used during training:

• learning_rate: 1e - 05
• train_batch_size: 4
• eval_batch_size: 4
• seed: 42
• optimizer: Adam with betas=(0.9,0.999) and epsilon = 1e - 08
• lr_scheduler_type: linear
• training_steps: 4000

Training results

Framework versions

• Transformers 4.20.0.dev0
• Pytorch 1.11.0+cu113
• Datasets 2.2.2
• Tokenizers 0.12.1

📄 License

This section is not provided in the original README, so it is skipped.