# Cutting the Error by Half: Investigation of Very Deep CNN and Advanced Training Strategies for Document Image Classification

arXiv: https://arxiv.org/abs/1704.03557 · HTML: https://mlsanity.com/paper/1704.03557/

## Key facts

- Cited by 22 papers in the map (data through 14 Oct 2026).
- First posted on arXiv 11 Apr 2017.

## Built on by (most cited first, 22 of 22)

- [DocFormer: End-to-End Transformer for Document Understanding](https://mlsanity.com/paper/2106.11539/) (2021)
- [DiT: Self-supervised Pre-training for Document Image Transformer](https://mlsanity.com/paper/2203.02378/) (2022)
- [LiLT: A Simple yet Effective Language-Independent Layout Transformer for Structured Document Understanding](https://mlsanity.com/paper/2202.13669/) (2022)
- [StructuralLM: Structural Pre-training for Form Understanding](https://mlsanity.com/paper/2105.11210/) (2021)
- [Document AI: Benchmarks, Models and Applications](https://mlsanity.com/paper/2111.08609/) (2021)
- [Bi-VLDoc: Bidirectional Vision-Language Modeling for Visually-Rich Document Understanding](https://mlsanity.com/paper/2206.13155/) (2022)
- [LayoutMask: Enhance Text-Layout Interaction in Multi-modal Pre-training for Document Understanding](https://mlsanity.com/paper/2305.18721/) (2023)
- [Evaluating Out-of-Distribution Performance on Document Image Classifiers](https://mlsanity.com/paper/2210.07448/) (2022)
- [VLCDoC: Vision-Language Contrastive Pre-Training Model for Cross-Modal Document Classification](https://mlsanity.com/paper/2205.12029/) (2022)
- [On Evaluation of Document Classification using RVL-CDIP](https://mlsanity.com/paper/2306.12550/) (2023)
- [EAML: Ensemble Self-Attention-based Mutual Learning Network for Document Image Classification](https://mlsanity.com/paper/2305.06923/) (2023)
- [DocXplain: A Novel Model-Agnostic Explainability Method for Document Image Classification](https://mlsanity.com/paper/2407.03830/) (2024)
- [Efficient Document Image Classification Using Region-Based Graph Neural Network](https://mlsanity.com/paper/2106.13802/) (2021)
- [Multimodal Side-Tuning for Document Classification](https://mlsanity.com/paper/2301.07502/) (2023)
- [Evaluating Adversarial Robustness on Document Image Classification](https://mlsanity.com/paper/2304.12486/) (2023)
- [DWT-CompCNN: Deep Image Classification Network for High Throughput JPEG 2000 Compressed Documents](https://mlsanity.com/paper/2306.01359/) (2023)
- [Image Generation and Learning Strategy for Deep Document Forgery Detection](https://mlsanity.com/paper/2311.03650/) (2023)
- [CICA: Content-Injected Contrastive Alignment for Zero-Shot Document Image Classification](https://mlsanity.com/paper/2405.03660/) (2024)
- [DP-DocLDM: Differentially Private Document Image Generation using Latent Diffusion Models](https://mlsanity.com/paper/2508.04208/) (2025)
- [Leveraging Semantic Segmentation Masks with Embeddings for Fine-Grained Form Classification](https://mlsanity.com/paper/2405.14162/) (2024)
- [WordVIS: A Color Worth A Thousand Words](https://mlsanity.com/paper/2412.10155/) (2024)
- [DocVCE: Diffusion-based Visual Counterfactual Explanations for Document Image Classification](https://mlsanity.com/paper/2508.04233/) (2025)

## Abstract (arXiv)

We present an exhaustive investigation of recent Deep Learning architectures, algorithms, and strategies for the task of document image classification to finally reduce the error by more than half. Existing approaches, such as the DeepDocClassifier, apply standard Convolutional Network architectures with transfer learning from the object recognition domain. The contribution of the paper is threefold: First, it investigates recently introduced very deep neural network architectures (GoogLeNet, VGG, ResNet) using transfer learning (from real images). Second, it proposes transfer learning from a huge set of document images, i.e. 400,000 documents. Third, it analyzes the impact of the amount of training data (document images) and other parameters to the classification abilities. We use two datasets, the Tobacco-3482 and the large-scale RVL-CDIP dataset. We achieve an accuracy of 91.13% for the Tobacco-3482 dataset while earlier approaches reach only 77.6%. Thus, a relative error reduction of more than 60% is achieved. For the large dataset RVL-CDIP, an accuracy of 90.97% is achieved, corresponding to a relative error reduction of 11.5%.