RFW0135: Explore SeamFormer pipeline for line segmentation
Summary
We need to explore SeamFormer pipeline for line segmentation and try to train with our labelled data.
Key Concepts
line segmentation: Process of detect line from an image of page.
SeamFormer pipeline: It is a line segmentation pipeline for handwritten documents.
Context
We have our own line segmentation model but it is not able to perform well. Hence we want to check with SeamFormer pipeline as they have trained their model on south Asian hand written manuscript. It has close match with Tibetan manuscript.
We need to prepare a bench mark data set to compare the approach.
Outputs
Comparative analysis report of both approaches using the benchmark data set. The report should consist:
accuracy
inference computational cost
Inputs
To explore and compare the line segmentation approaches, you will be given the line segmented labelled data and the existing model.
Timeline
Specify the expected delivery date for the project.
RFW0135: Explore SeamFormer pipeline for line segmentation
Summary
We need to explore SeamFormer pipeline for line segmentation and try to train with our labelled data.
Key Concepts
line segmentation: Process of detect line from an image of page. SeamFormer pipeline: It is a line segmentation pipeline for handwritten documents.
Context
We have our own line segmentation model but it is not able to perform well. Hence we want to check with SeamFormer pipeline as they have trained their model on south Asian hand written manuscript. It has close match with Tibetan manuscript. We need to prepare a bench mark data set to compare the approach.
Outputs
Comparative analysis report of both approaches using the benchmark data set. The report should consist:
Inputs
To explore and compare the line segmentation approaches, you will be given the line segmented labelled data and the existing model.
Timeline
Specify the expected delivery date for the project.
References
SeamFormer pipeline research paper.