Hi, I found your repo by coincidence and am very interseted in the your proposed multitask framework. I noticed that you only the output of cnn model to do a 81 multi-label prediction. I assumed that it is for better learning of image representation. However, for you did not attach your results or model's performance, I really wonder how it work.
Looking forward to hearing from you, thanks.
Hi, I found your repo by coincidence and am very interseted in the your proposed multitask framework. I noticed that you only the output of cnn model to do a 81 multi-label prediction. I assumed that it is for better learning of image representation. However, for you did not attach your results or model's performance, I really wonder how it work.
Looking forward to hearing from you, thanks.