GuideWsp
GuideWsp
before the data was sended the net, it was transformed to 0-1. at the begin of training, all the data was sended to the net, but i found that the...
the alpha matte is between 0-1. and the input have transform as vggnet.
We have make alpha by ourselves, but we can't get similar result. If possible, please give some advise.
it is wrong. when you scale the prob, so after clip, all the prob are zero. i have check it. so i am sure it was wrong
I change it as : prob = F.softmax(score, axis=1) prob = prob - 0.5 prob = F.clip(prob, -0.5, 0.5) prob = prob + 0.5 finaly, i have to abandon this...
you can try "Deep Image Matting". if you give a relative good trimap, it will give a good results. but it isn't an automatic algorithm. i nearly give up this...
maybe it's right. can you tell me when you get the results as the paper shown? thanks
I had tried rife-HD and rife-anime model, all can't get the right result.
I run the nanodet with the same ncnn.lib+ncnn.dll, the result is right.
due to limited gpu, I just try the vip_s7, until to now, the accuracy of vip_s7 is 84.68%。 