Closed Zigars closed 2 years ago
basically, you just need to convert the output to a numpy array, then draw a heatmap with matplotlib
Actually, the anomaly_map's datas are like this:
I don't know how to caculate this anomaly_map from a normalization flow, and the original paper doesn't mention how to get the anomaly_map, even loss function haven't mention either. I'm confuse about these problem.
if this is your anomaly map then each value is the anomaly score for each pixel, just draw a heatmap of it
I want use your code in my own dataset, but I do not know how to visualization the anomaly map. So I can't see the detection results derectly. Can you give me some advice?