I used the original probs for Place 365 dataset ,the colored picture have many discontinous color blocks.
I referred the paper to create new ab prior_probs of Place365 dataset,but the colored picture are most green and blue ,the result is worse than using the original probs .I dont know what's wrong with my new probs? Can you show a code for the prior_probs?Thanks a lot!
I used the original probs for Place 365 dataset ,the colored picture have many discontinous color blocks. I referred the paper to create new ab prior_probs of Place365 dataset,but the colored picture are most green and blue ,the result is worse than using the original probs .I dont know what's wrong with my new probs? Can you show a code for the prior_probs?Thanks a lot!