Table Transformer (TATR) is a deep learning model for extracting tables from unstructured documents (PDFs and images). This is also the official repository for the PubTables-1M dataset and GriTS evaluation metric.
I am using below model configuration for Table Detection
Model | Training Data | Model Card | File | Size
DETR R18 | PubTables-1M | Model Card | Weights | 110 MB
The weights are pretty good for table detection but I see some cases where it detect non trabular data.
Immediate solution came to my mind is finetune the model with negative samples.
While figuring it out I bumped up into this #113
Anything I am missing here ? What else can be probable solutions @bsmock
Hello Guys,
I am using below model configuration for Table Detection
Model | Training Data | Model Card | File | Size DETR R18 | PubTables-1M | Model Card | Weights | 110 MB
The weights are pretty good for table detection but I see some cases where it detect non trabular data. Immediate solution came to my mind is finetune the model with negative samples. While figuring it out I bumped up into this #113
Anything I am missing here ? What else can be probable solutions @bsmock