CascadeTabNet:An Approach for end to end table detection and structure recognition from image-based documents(2020)
Paper
DevashishPrasad/CascadeTabNet
- Network structure
- Iterative transfer learning
uses multi-task learning methods and multiple iterations of transfer learning to achieve goals. First train the model on a general data set, and then fine-tune multiple times for a specific data set.
- Image Transformation and data augmentation (image transformation and data augmentation)
- Dilation transform
In the dilation transform, the original image is first converted into a binary image, and then the original image is transformed to thicken the black pixel area. - Smudge transform
- Dilation transform
- Pipeline
- result