huggingface/pytorch-image-models · warning

Transform returned None for index {idx}. Skipping sample.

Error message

Transform returned None for index {idx}. Skipping sample.

What it means

NaFlexDataset.__iter__ applies the per-sample transform and checks for None returns. A None means the transform pipeline explicitly returned None (common with 'bad image' guards like PIL's Image.open failure fallbacks or timm's image-not-loaded checks) and the sample is skipped with a warning.

Source

Thrown at timm/data/naflex_dataset.py:544

                 continue

            # Get the pre-initialized transform and patchifier using patch_idx
            transform_key = (seq_len, patch_idx)
            transform = self.transforms.get(transform_key)
            batch_patchifier = self.patchifiers[patch_idx]

            batch_imgs = []
            batch_targets = []
            for idx in indices:
                try:
                    # Get original image and label from map-style dataset
                    img, label = self.base_dataset[idx]

                    # Apply transform if available
                    # Handle cases where transform might return None or fail
                    processed_img = transform(img) if transform else img
                    if processed_img is None:
                        warnings.warn(f"Transform returned None for index {idx}. Skipping sample.")
                        continue

                    batch_imgs.append(processed_img)
                    batch_targets.append(label)

                except IndexError:
                     warnings.warn(f"IndexError encountered for index {idx} (possibly due to padding/repeated indices). Skipping sample.")
                     continue
                except Exception as e:
                    # Log other potential errors during data loading/processing
                    warnings.warn(f"Error processing sample index {idx}. Error: {e}. Skipping sample.")
                    continue # Skip problematic sample

            if self.mixup_fn is not None:
                batch_imgs, batch_targets = self.mixup_fn(batch_imgs, batch_targets)

            batch_imgs = [batch_patchifier(img) for img in batch_imgs]
            batch_samples = list(zip(batch_imgs, batch_targets))

View on GitHub (pinned to 9a5261e31b)

Solutions

1. Audit your transform chain: every branch must return a tensor; a missing return yields None
2. Remove or repair the flagged dataset samples
3. If a guard transform intentionally returns None for bad images, accept the skip and clean the dataset

Example fix

# before
class MyTransform:
    def __call__(self, img):
        if img is None:
            return None  # or: forgets return on some path
# after
class MyTransform:
    def __call__(self, img):
        if img is None:
            raise ValueError('bad image')  # or return a placeholder tensor
        return do_process(img)

out = transform(sample_img)\nassert out is not None, 'transform pipeline has a None-returning branch'

When it happens

Trigger: A transform whose failure mode is returning None — e.g. a custom transform returning None on decode failure, or timm's ImageNetInfo/bad-image transforms — hitting a corrupt or missing sample during iteration.

Common situations: Datasets with unreadable entries where the transform authors chose None over raising; also custom user transforms that forget to return on some branch, making every sample return None (then warnings flood and batches are empty).

Related errors

• Error processing sample index {idx}. Error: {e}. Skipping sa
• {name.capitalize()} range reversed. Swapping.
• final_scale_range values should ideally be between 0.0 and 1
• Final scale randomization ({scale_factor:.2f}) resulted in s
• Input image must have positive dimensions, got H={height}, W

AI-assisted analysis of huggingface/pytorch-image-models@9a5261e31b (2026-08-27). Data as JSON: /api/errors/2cd2db1d87f35109. Report an issue: GitHub.