DocStruct: A Multimodal Method to Extract Hierarchy Structure in Document for General Form Understanding.
Zilong WangMingjie ZhanXuebo LiuDing LiangPublished in: CoRR (2020)
Keyphrases
- experimental evaluation
- high accuracy
- significant improvement
- multiscale
- neural network
- special case
- high precision
- clustering method
- computational complexity
- objective function
- multimedia
- prior knowledge
- cost function
- image segmentation
- semi supervised
- feature set
- mutual information
- multi modal
- document collections
- feature selection
- information retrieval