NVIDIA / NVIDIA/DALI

Apply DALI operations (e.g. pad, normalize) to the entire dataset instead of per batch / per image

Open
#3,127 8 comments 0 reactions 0 assignees View on GitHub

Nobody has claimed this yet.

question
Dominant language
C++
Stars
5.8k
Forks
678
Avg merge
3d 1h
Merged PRs (30d)
27

Description

Is it possible to apply DALI operations to the entire dataset instead of per batch / per image? For example, I want to apply the same padding operation all the images in a dataset to increase their size from 32x32 to 40x40 and then train a model using the padded dataset. In this use case, I think repeatedly performing the same operation to all images in a dataset for a high number of epochs may generate unnecessary overhead. Thank you.

Contributor guide

Open the contributing guide

First steps

  1. Read the whole issue, then the project's contributing guide.
  2. Comment on the issue to say you are picking it up — it saves two people doing the same work.
  3. Fork the repository and make your change on a branch.
  4. Open a pull request that references the issue number.

Research direction

Start by tracing how DALI operations are executed per batch or per image and how a dataset is represented across epochs; the issue names no files or tests to begin with. Determine whether dataset-wide padding or normalization can be supported without changing existing per-batch behavior, and define completion around avoiding repeated work while preserving training results.

Written by the indexing model from the issue text.

Assessment

Tech stack
cpp
Domain
data, machine-learning
Issue type
Feature
Difficulty
5/5
Estimated time
Over a week
Activity status
Stale
Clarity
Needs clarification
Newbie friendliness
25/100

Get new issues in your inbox

A short digest of beginner-friendly GitHub issues.