Instance segmentation is performed on the whole image over five different classes. The evaluation is performed according to the COCO evaluation metric. We use the mean average precision (mAP) over different intersection over union (IoU) thresholds, namely 0.50:0.05:0.95 (primary COCO challenge metric), and denote this metric by AP. We also report the mAP over an IoU of 0.50 (PASCAL VOC metric) and 0.75 (strict metric), denoted as AP50 and AP75, respectively. The evaluation metrics for instance segmentation are exactly the same as for object detection, except that the mean intersection over union is calculated over the masks instead of the boxes. We use the official pycocotools functions to calculate the performance.

NameTrain CarAPAP50AP75AP (per class)AP50 (per class)AP75 (per class)PaperCodeRGBGrayDepthAdditionalTeamTitleConference
SVIRO-TeamX50.2580.520.221 IS: 0.33    CS: 0.23    Person: 0.24    Object: 0.23 IS: 0.78    CS: 0.38    Person: 0.60    Object: 0.32 IS: 0.25    CS: 0.27    Person: 0.09    Object: 0.28NoYesNoYes