Albumentations to DALI Transform Mapping
On this page
This page maps Albumentations transforms to the closest DALI operation when DALI has a practical direct equivalent. Use it as a capability guide, not as a performance page.
How to Read This Page
- A named DALI operation means DALI has a built-in graph operation for the same broad operation.
-means DALI does not support that transform as a built-in graph augmentation primitive.- The table shows built-in DALI operations. Custom DALI graph code is not counted as DALI support.
- Unsupported rows stay visible because transform coverage is part of the library choice.
Execution Model
- Albumentations receives an already decoded OpenCV-style
H,W,Cchannel-last array and applies a Python augmentation pipeline to one sample and its targets. - DALI builds a data pipeline graph that can include reading, mixed-device image decoding, augmentation, normalization, output layout conversion, prefetching, and framework handoff.
- DALI operators define graph nodes and symbolic data nodes. They are not direct per-sample Python augmentation calls.
- DALI pipeline performance depends on the whole graph: decode, transfer scope, prefetching, framework handoff, and GPU memory.
Direct DALI Mappings
| Albumentations transform | DALI operation |
|---|---|
Resize | resize |
RandomCrop | resize plus crop |
RandomResizedCrop | random_resized_crop |
HorizontalFlip | flip |
VerticalFlip | flip |
Pad | pad |
Rotate | rotate |
Affine | warp_affine |
RandomBrightnessContrast | brightness_contrast |
ColorJitter | color_twist |
HueSaturationValue | hue and saturation |
CLAHE | clahe |
Equalize | equalize |
GaussianBlur | gaussian_blur |
GaussNoise | noise.gaussian |
SaltAndPepper | noise.salt_and_pepper |
Erasing | erase |
ImageCompression | jpeg_compression_distortion |
Normalize | crop_mirror_normalize |
No Direct DALI Mapping
These categories are the practical reason DALI should not be presented as a general Albumentations replacement:
| Category | Examples | Why it matters |
|---|---|---|
| Non-rigid geometry beyond the mapped subset | ElasticTransform, GridDistortion, ThinPlateSpline, OpticalDistortion | These policies are common for robustness testing and domain-specific augmentation. |
| Weather, camera, and illumination effects outside simple color changes | RandomRain, RandomSnow, AtmosphericFog, LensFlare, PlasmaShadow | These are augmentation policies, not only graph preprocessing steps. |
| Dropout and object-aware policies | CoarseDropout, GridDropout, ConstrainedCoarseDropout, CopyAndPaste | These often need target-aware bookkeeping, not only pixel operations. |
| Detection-safe crops and multi-image policies | BBoxSafeRandomCrop, RandomSizedBBoxSafeCrop, Mosaic | Albumentations exposes these as augmentation policies with target contracts. |
| OBB and keypoint policies | OBB-safe transforms, keypoint visibility/filtering workflows | DALI does not provide the same high-level target-aware policy contract. |
| Volumetric and broad non-RGB workflows | RandomCrop3D, CubicSymmetry, CoarseDropout3D | Do not generalize RGB image operators to volume or arbitrary-channel training. |