A repository contains the code for various semantic segmentation in TensorFlow and PyTorch framework.
I watched your tutorial on YouTube, and it was so interesting but am having problem with categorical dataset even after changing 1 to 7, binary_cross _entropy to categorical_cross_entropy, sigmoid to softmax. Please help me out.
Hi, First, thank you for sharing your deeplabv3 code. When I try to run the code, I get a backbone error in Resnet 50. Like there is no layer such as 'conv4_block6_out' and 'conv2_block2_out'. Despite my research, I could not solve this problem. I will be glad if you help. Best Wishes,
hello, can you check my model below. i get wrong segmentation output. i have used upsampling2d instead of conv2dtranspose. also i used number of filters in decoder block as multiples of 64. whether it should match the encoder filters?
edit: it works, but if we increase the number of filters will it give good performance? i got slightly more accuarcy, but the mean iou seems to be less.