Preprocesses a single image Returns: crops: (n_crops, n_patches, patch_dim) individual crops, `n_crops` might change between images but the other dimension are fixed tokens: (n_tokens,) int32 tokens, pad tokens indicate where to insert the
(self, image, is_training: bool, rng=None)
| 594 | return image_input_idx |
| 595 | |
| 596 | def preprocess(self, image, is_training: bool, rng=None): |
| 597 | """Preprocesses a single image |
| 598 | |
| 599 | Returns: |
| 600 | crops: (n_crops, n_patches, patch_dim) individual crops, `n_crops` might |
| 601 | change between images but the other dimension are fixed |
| 602 | tokens: (n_tokens,) int32 tokens, pad tokens indicate where to insert the |
| 603 | patch features, might include other special tokens as well |
| 604 | image_idx: (n_crops, n_patches) index in `tokens` to put the patch features from the |
| 605 | crops after pooling, negative values indicates patches features to exclude |
| 606 | padding_mask: (n_crops, n_patches) what percent of each crop is padding, can be None |
| 607 | if the image mask is not being used. |
| 608 | """ |
| 609 | crops, image_tokens, patch_ordering, img_mask = self.image_to_patches_and_tokens( |
| 610 | image, is_training, rng) |
| 611 | patch_idx = self.build_image_input_idx( |
| 612 | image_tokens, |
| 613 | patch_ordering, |
| 614 | ) |
| 615 | return crops, image_tokens, patch_idx, img_mask |
| 616 | |
| 617 | def __call__( |
| 618 | self, |
no test coverage detected