# Learning to Sample the Most Useful Training Patches from Images

arXiv: https://arxiv.org/abs/2011.12097 · HTML: https://mlsanity.com/paper/2011.12097/

## Key facts

- Cited by 4 papers in the map (data through 14 Oct 2026).
- First posted on arXiv 24 Nov 2020.

## Built on by (most cited first, 4 of 4)

- [Semi-Supervised Semantic Segmentation with Pixel-Level Contrastive Learning from a Class-wise Memory Bank](https://mlsanity.com/paper/2104.13415/) (2021)
- [TransMix: Attend to Mix for Vision Transformers](https://mlsanity.com/paper/2111.09833/) (2021)
- [SamplingAug: On the Importance of Patch Sampling Augmentation for Single Image Super-Resolution](https://mlsanity.com/paper/2111.15185/) (2021)
- [ISP meets Deep Learning: A Survey on Deep Learning Methods for Image Signal Processing](https://mlsanity.com/paper/2305.11994/) (2023)

## Abstract (arXiv)

Some image restoration tasks like demosaicing require difficult training samples to learn effective models. Existing methods attempt to address this data training problem by manually collecting a new training dataset that contains adequate hard samples, however, there are still hard and simple areas even within one single image. In this paper, we present a data-driven approach called PatchNet that learns to select the most useful patches from an image to construct a new training set instead of manual or random selection. We show that our simple idea automatically selects informative samples out from a large-scale dataset, leading to a surprising 2.35dB generalisation gain in terms of PSNR. In addition to its remarkable effectiveness, PatchNet is also resource-friendly as it is applied only during training and therefore does not require any additional computational cost during inference.