Open tommying opened 8 months ago
Hi, thanks for contributing a good work.
Whether an augmentation image is required as input to model.encode_image.
The paper mentioned that 'the visual encoder without any image augmentation,' but only achieves the performance stated in the paper when image augmentation is used as input.
the resluts with image augmentation:
the resluts without image augmentation:
Hi, thanks for contributing a good work.
Whether an augmentation image is required as input to model.encode_image.
The paper mentioned that 'the visual encoder without any image augmentation,' but only achieves the performance stated in the paper when image augmentation is used as input.
the resluts with image augmentation:
the resluts without image augmentation: