Many incorrectly labeled images #8
Comments
|
Excuse me， may I ask you that how you get the FairFace dataset? I found that I can not find the download link of FairFace. |
https://drive.google.com/file/d/1Z1RqRo0_JiavaZw2yzZG6WETdZQ8qX86/view if you scroll to the bottom of the readme the link for this dataset (padding 0.25) and another are available Best of luck on your endeavor! |
|
The annotations may not be perfect due to the subjective nature of the task. The paper is about perceived, not self-identified, race (and also gender, age), like most existing face attribute classification papers/datasets. |
|
Excuse me，may I ask you that how you get the pretrained models? I could not find the download link of the pretrained models. |
The paper reads "We further refined the annotations by training a model from the initial ground truth annotations and applying back to the dataset. We then manually re-verified the annotations for images whose annotations differ from model predictions."
But I tried to do this check myself, and I found many examples of annotations that do not match the model predictions. I do not have an estimate for how many images have this problem. But here are a few samples from the validation dataset that are annotated as "White", along with the model's predictions for these images. I would guess 2-3 of these 64 people would self-identify as "White":
The text was updated successfully, but these errors were encountered: