# Creating comparable pipeline with DALI

**URL:** <https://forums.developer.nvidia.com/t/creating-comparable-pipeline-with-dali/263226>\
**Category:** Frameworks (archived)\
**Created:** [August 15, 2023, 6:43pm UTC](https://forums.developer.nvidia.com/t/creating-comparable-pipeline-with-dali/263226 "2023-08-15T18:43:45Z")\
**Posts on this page:** 1\
**Page:** 1

<div class="post-metadata">

**Author:** ![thanks](https://developer.download.nvidia.com/images/forums/profile-default-devtalk-84.png) [@thanks](https://forums.developer.nvidia.com/u/thanks)\
**Post date:** [August 15, 2023, 6:43pm UTC](https://forums.developer.nvidia.com/t/creating-comparable-pipeline-with-dali/263226/1 "2023-08-15T18:43:45Z")

</div>

Hey there everybody!

I’m super new to this space, and I’m just starting to explore DALI as a means of speeding up my image classification model training. I’m realizing some pretty serious performance improvements, but I’m having trouble replicating the “accuracy” of my model that uses TorchVision to transform the images in my pipeline.

Does anyone know of a good resource that I could use to (more or less) replicate the following TorchVision pipeline using DALI?

```auto
transform = transforms.Compose([
    transforms.Resize((224, 224)),
    transforms.RandomHorizontalFlip(p=0.5),
    transforms.ColorJitter(brightness=0.2, contrast=0.2, saturation=0.2, hue=0.2),
    transforms.RandomAffine(degrees=30, translate=(0.1, 0.1), scale=(0.8, 1.2), shear=10),
    transforms.ToTensor(),
    transforms.Normalize(mean=[0.485, 0.456, 0.406], std=[0.229, 0.224, 0.225])
])

```

Thanks everyone!