Are Vision Language Models Texture or Shape Biased and Can We Steer Them?
Paper
• 2403.09193 • Published
• 9
Datasets:
image imagewidth (px) 224 224 | lf_label stringclasses 16 values | hf_label stringclasses 16 values |
|---|---|---|
airplane | airplane | |
airplane | airplane | |
airplane | airplane | |
airplane | airplane | |
airplane | airplane | |
airplane | bear | |
airplane | bear | |
airplane | bear | |
airplane | bear | |
airplane | bear | |
airplane | bicycle | |
airplane | bicycle | |
airplane | bicycle | |
airplane | bicycle | |
airplane | bicycle | |
airplane | bird | |
airplane | bird | |
airplane | bird | |
airplane | bird | |
airplane | bird | |
airplane | boat | |
airplane | boat | |
airplane | boat | |
airplane | boat | |
airplane | boat | |
airplane | bottle | |
airplane | bottle | |
airplane | bottle | |
airplane | bottle | |
airplane | bottle | |
airplane | car | |
airplane | car | |
airplane | car | |
airplane | car | |
airplane | car | |
airplane | cat | |
airplane | cat | |
airplane | cat | |
airplane | cat | |
airplane | cat | |
airplane | chair | |
airplane | chair | |
airplane | chair | |
airplane | chair | |
airplane | chair | |
airplane | clock | |
airplane | clock | |
airplane | clock | |
airplane | clock | |
airplane | clock | |
airplane | dog | |
airplane | dog | |
airplane | dog | |
airplane | dog | |
airplane | dog | |
airplane | elephant | |
airplane | elephant | |
airplane | elephant | |
airplane | elephant | |
airplane | elephant | |
airplane | keyboard | |
airplane | keyboard | |
airplane | keyboard | |
airplane | keyboard | |
airplane | keyboard | |
airplane | knife | |
airplane | knife | |
airplane | knife | |
airplane | knife | |
airplane | knife | |
airplane | oven | |
airplane | oven | |
airplane | oven | |
airplane | oven | |
airplane | oven | |
airplane | truck | |
airplane | truck | |
airplane | truck | |
airplane | truck | |
airplane | truck | |
bear | airplane | |
bear | airplane | |
bear | airplane | |
bear | airplane | |
bear | airplane | |
bear | bear | |
bear | bear | |
bear | bear | |
bear | bear | |
bear | bear | |
bear | bicycle | |
bear | bicycle | |
bear | bicycle | |
bear | bicycle | |
bear | bicycle | |
bear | bird | |
bear | bird | |
bear | bird | |
bear | bird | |
bear | bird |
End of preview. Expand
in Data Studio
This dataset contains images of natural objects with low- and high-frequency cue conflicts, as described in the paper Can We Talk Models Into Seeing the World Differently?.
The dataset consists of 1200 images where the low-frequency and high-frequency information correlate with different classes (out of 16 classes). This dataset can be used to evaluate the frequency-bias of vision models.
- Downloads last month
- 22
Size of downloaded dataset files:
43.5 MB
Size of the auto-converted Parquet files:
43.5 MB
Number of rows:
1,280