In this paper, we proposed a BB generator and a positive RoI generator. We showed that generated RoIs can be used both as an analysis tool (owing to its controllable nature) and a training method for the two-stage object detectors.
In this work, we have shown the benefits of using fine-tuning from a close domain in the context of table detection. In addition to the accuracy improvement, this approach avoids overfitting and solves the problem of having a small dataset. Moreover…
There aren't more papers