Hi all,

At the request of some teams, we are happy to provide some basic statistics about the preliminary testing set. Please keep in mind that this testing set is designed to ensure that your algorithm can be evaluated correctly, and is not designed to be representative of your algorithm's true performance on the final testing set. The preliminary set contains a total of 16 videos, parsed at 1fps for a total of 4897 images. The tools present, and the number of frames that each tool is present in are as follows:

{'bipolar forceps': 2500,
 'cadiere forceps': 1517,
 'monopolar curved scissor': 2660,
 'force bipolar': 103,
 'needle driver': 2474,
 'clip applier': 68,
 'grasping retractor': 68}

We are not able to provide AP scores for each class as feedback.

Thanks,

SurgToolLoc Organizing Committee