All participating teams will automatically be considered for the

Pallas Ludens Endoscopic Vision Award

sponsored by PALLAS LUDENS (http://pallas-ludens.com).

The categories for obtaining awards for this challenge are the following:

1st Category: Polyp Localization Award

The objective of this category is to check whether your method is able to locate correctly the polyp in a given image in a way such ONLY the area containing the polyp is marked/highlighted. The output of a given method should provide, for any given image, a series of potential polyp localizations (preferably one per polyp).

The metrics for this category will be the following:

  • Localization score (recall): Loc-score measured as Loc-score = TP/(TP+FN).
  • Localization precision: Loc-prec, measured as Loc-prec = TP/(TP+FP).
  • (TP, FP and FN) are defined in http://polyp.grand-challenge.org/metrics/

The database used for this category will be ETIS-Larib.

Awards:

  • Precision award (100 €) to the team with highest precision score
  • Localization award (100 €) to the team with highest localization score (less polyps missed)

2nd Category: Polyp Detection Award

The objective of this category is to check whether your method is able to detect correctly polyp presence (and absence) in a video sequence which may contain or not a polyp. The output of a given method should provide, for any given image, a series of potential polyp localizations (preferably one per polyp) in a given frame.

The metrics for this category will be the following:

  • Overall F-score: F-score = 2*(precision*recall)/ (precision + recall) where precision = TP/(TP+FP) and recall = TP/(TP+FN).
  • Average F-score: Average F-score value across all the videos.
  • Detection latency (DL): DL = (# of frame with first detection - # of frame with first polyp apparition)/frame_rate.

The database used for this category will be ASU-Mayo.

Awards:

  • Overall F-Score (100 €) to the team with highest overall F-score
  • Average F-Score (100 €) to the team with highest average F-score
  • Detection Latency Award (100 €) to the team with lowest detection latency.