Fine-tuned convolutional neural networks for feature extraction and classification of scanned document images using semi-automatic labelling approach
Bibliographic record
Abstract
Organising documents into relevant categories through image classification is crucial for management and safeguarding of valuable information. Many studies have done work on it with manual intervention, but still there is a scope of improvement. After finding gaps in existing studies, this research fine-tuned a hyper-parameter of pre-trained model based on various convolutional neural networks (CNNs), specifically the EfficientNetB3 and DenseNet201 models, for feature extraction and classification. These models are fine-tuned with the subset of the Ryerson Vision Lab Complex Document Information Processing (RVL_CDIP) dataset. The dataset comprises 16,000 image-scanned documents categorised into 16 classes with semi-automatic approach of labelling. The modified models are fine-tuned by adding a few more layers. The modified models outperformed in terms of accuracy, precision, recall and F1-Score for EfficientNetB3 and DenseNet201. These results highlight a significant improvement when comparing the proposed CNN models with baseline models through the utilisation of semi-automatic labelling and fine-tuning.
Fetched live from OpenAlex and de-inverted. Abstracts are not stored in this database: the inverted indexes are 8.6 GB of the frame’s 9.3 GB of text, and the host has 13 GB free.
How this classification was reachedexpand
Full frame machine prediction
Teacher imitationNot calibrated prevalence, not ground truth. Human validation pending. The Gemma side is a direct model label for every work in the frame, read from the title-only record. The Codex side is a classifier learned from the 10,348 direct Codex labels and calibrated to design-weighted sample rates; fields without enough sample support carry no Codex call. Candidate is the union of the two sides; consensus is their intersection. These outputs are machine_predicted_unvalidated and are not human labels.
Distilled classifier scores by category (both heads)
| Category | Codex | Gemma |
|---|---|---|
| Metaresearch | 0.001 | 0.002 |
| Meta-epidemiology (narrow) | 0.001 | 0.000 |
| Meta-epidemiology (broad) | 0.001 | 0.001 |
| Bibliometrics | 0.001 | 0.001 |
| Science and technology studies | 0.000 | 0.000 |
| Scholarly communication | 0.001 | 0.001 |
| Open science | 0.002 | 0.001 |
| Research integrity | 0.001 | 0.001 |
| Insufficient payload (model declined to judge) | 0.001 | 0.001 |
Machine scores (provisional)
The two teacher heads of the student model, read on this work. A score orders the frame for review; it never asserts a category, and the validation status ships verbatim with every row.
Baseline scores from an immature model (maturity gate not passed, 7 training rounds). Scores rank; they never assert a category.
score_only:v0-immature-baseline · verbatim from the scoring run: score_only means the number may rank works, and no category label ships from itClassification
machine, unvalidatedMachine predicted; a candidate call from one source (direct Gemma or distilled Codex), not a consensus.
How this classification was reached, model by model and score by score, is at the end of the page under "How this classification was reached".