Event-based Object Segmentation

Based on the MVSEC dataset, we select some image-event pairs to evaluate the segmentation performance, namely MVSEC-SEG, which only serves as a test set. The MVSEC-SEG consists of 500 image-event pairs each in ”indoor flying1” and ”outdoor day2” sequences, containing 54,600 masks.

To perform universal event stream segmentation, we collected a large-scale RGB-Event dataset for event-centric segmentation, from current available pixel-level aligned datasets (VisEvent, COESOT), namely RGBE-SEG. The RGBE-SEG included 65,957 image-event pairs, 64,957 for training and 1,000 for testing. The test set contained 38,760 masks, and we artificially divided it into easy, medium, and hard subsets based on the complexity of scenarios. All ground truth masks were generated by images and the well-trained SAM.

Based on the DDD17 dataset, we select some image-event pairs to evaluate the segmentation performance, namely DDD17-SEG, which only serves as a test set. The DDD17-SEG consists of 1,000 image-event pairs in five sequences (dir0,dir1,dir3,dir4,dir7), containing 86,000 masks.

Based on the DSEC dataset, we select some image-event pairs to evaluate the segmentation performance, namely DSEC-SEG, which only serves as a test set. The DSEC-SEG consists of 1,000 image-event pairs in two sequences (zurich_city_01_a, zurich_city_04_c), containing 63,200 masks.