First, I want to thank you for your amazing work. I have some questions regarding the Retail50K dataset
- I have downloaded the Retail50K dataset and tried visualizing some images with polygon labels. I noticed that there are some potential missing annotations. For example, ID 122645 has a bottom row with price tags but that row is not annotated. I think missing annotation can lead to high False Negative of the trained model, and there is quite a lot of missing annotation cases like this. Can you verfify some missing annotations?
- In the csv files I noticed that the field 'ProductId' has 2 possible values: 1 and 1000050. After visualizing some samples, I assume that 1 means "row" and 1000050 means "nearby shelves". Is this correct?
First, I want to thank you for your amazing work. I have some questions regarding the Retail50K dataset