Visualizing the predicated Image


#1

I am getting validating accuracy of 97% but when I am predicting I am not able to visualize that image. what should I do?


#2

Be careful of the metric that you use. This data is extremely unbalanced in that class 0 is much more common than classes 1 or 2, therefore a “null classifier” that predicts only 0 will achieve high accuracy while still performing poorly. The metric we will be using to rank participants is average Kidney and Tumor Sørensen–Dice Score.