Closed nottombrown closed 5 years ago
How did you find these images?
@carlini - These bad images were due to the issue with the tasker pipeline. I filed this as an incident and will directly export data from the tasker database myself in the future.
We had been aware of the issue earlier, but I had failed to commit my changes that used the updated csv file.
I have manually verified that all the images in the following dataset folders appear unambiguous to me and are not truncated:
@carlini can I get a quick review?
Let me take a look
It all seems fine
This PR does the following:
train
andtest
dataset to exclude images that are truncated. Note that this reduces the size of the IID train and test sets to 125 images each. However, the excluded images can still be used for training in the largeextras
datasetResearchers can upgrade to the new dataset by pulling the repo and running the following command
Bad bird datapoints that are now removed
Bad bicycle datapoints that are now removed