Hi! Great work with impressive performance. I notice that not all VisEvent datasets are used for training, as stated, "Due to
the incomplete raw event data, 295 sequences are used for
training and 219 for testing in our setting."
Could you provide the selected sequence list? Thanks.