The evaluation is performed by comparing the area of the detected text region with the area of the ground truth area.

TASKS

You can submit results for each of the following tasks:

·   Task 1 (Batch Learning): Given all the available training examples, learn a model to detect text regions.

·   Task 2 (Online Learning): Examine the effect of limited training resources on the learning process by incrementally adding (in the same order as listed in the Model Selection Set file) the provided examples to the training data.

·   Task 3 (Active Learning): Examine the effect of selecting which examples to add to the training data.