aiqbench analysis
SWITi addresses a critical inference challenge in large-scale computer vision by mitigating stitching artifacts inherent in tiled neural network predictions. This method improves reconstruction fidelity for posterior sampling models without requiring additional forward passes, offering significant practical value for biomedical imaging and other domains relying on high-resolution image synthesis or segmentation. The introduction of reference-free metrics like FRT and ASV provides a robust framework for quantifying seam quality directly from pixel gradients.
SWITi: Quantifying and Reducing Tiling Artifacts with Sliding Window Inner Tiling
arXiv:2607.18990v1 Announce Type: cross Abstract: SWITi is a test-time method for reducing artifacts in tiled predictions, particularly for neural networks that learn posterior distributions from which solutions are sampled at inference time. Tiled predictions are unavoidable for large image data, and artifacts arise whenever tiles are smaller than a network's receptive field and when tiles are independent posterior samples. SWITi averages overlapping sliding-window predictions, so discrepancies between neighboring samples are spread across shifted tile positions rather than accumulating at fixed seam coordinates. For posterior models, SWITi uses no more tile samples than an MMSE estimate requires and therefore incurs no additional forward passes. Additionally, we introduce two reference-free metrics, the Fraction of Rejected Tests (FRT) and Artifact Severity (ASV), for detecting and quantifying tiling artifacts from a per-tile permutation test that compares the distribution of pixel gradients across tile seams against the surrounding image content. On pre-trained and published image splitting models across three fluorescence microscopy datasets in 2D and 3D, we show that SWITi substantially attenuates stitching seams while also improving reconstruction fidelity and resolution. Since tiling artifacts in posterior predictions can easily be mistaken for biological structures or for boundaries between biological structures, removing or reducing them using SWITi will improve the downstream processing of large image predictions, which is particularly relevant for biomedical data.
Abstract:SWITi is a test-time method for reducing artifacts in tiled predictions, particularly for neural networks that learn posterior distributions from which solutions are sampled at inference time. Tiled predictions are unavoidable for large image data, and artifacts arise whenever tiles are smaller than a network's receptive field and when tiles are independent posterior samples. SWITi averages overlapping sliding-window predictions, so discrepancies between neighboring samples are spread across shifted tile positions rather than accumulating at fixed seam coordinates. For posterior models, SWITi uses no more tile samples than an MMSE estimate requires and therefore incurs no additional forward passes. Additionally, we introduce two reference-free metrics, the Fraction of Rejected Tests (FRT) and Artifact Severity (ASV), for detecting and quantifying tiling artifacts from a per-tile permutation test that compares the distribution of pixel gradients across tile seams against the surrounding image content. On pre-trained and published image splitting models across three fluorescence microscopy datasets in 2D and 3D, we show that SWITi substantially attenuates stitching seams while also improving reconstruction fidelity and resolution. Since tiling artifacts in posterior predictions can easily be mistaken for biological structures or for boundaries between biological structures, removing or reducing them using SWITi will improve the downstream processing of large image predictions, which is particularly relevant for biomedical data.
Computer Vision and Pattern Recognition (cs.CV); Artificial Intelligence (cs.AI)
arXiv:2607.18990 [cs.CV]
(or arXiv:2607.18990v1 [cs.CV] for this version)
https://doi.org/10.48550/arXiv.2607.18990 Focus to learn more arXiv-issued DOI via DataCite (pending registration)
arXivLabs is a framework that allows collaborators to develop and share new arXiv features directly on our website.
Both individuals and organizations that work with arXivLabs have embraced and accepted our values of openness, community, excellence, and user data privacy. arXiv is committed to these values and only works with partners that adhere to them.
Have an idea for a project that will add value for arXiv's community? Learn more about arXivLabs.
Full text under the publisher's open licence · 312 words
Related coverage
Co-Designing AI Model Attention for Fast, Interactive Long-Context Inference
NVIDIA · Jul 31, 2026
Game Design Creativity evals
LessWrong · Oct 12, 2026
Productionizing TurboQuant on AMD GPUs for KV-Cache-Bound LLM Inference
AMD · Jun 11, 2026
[AINews] High Taste Testers Ep. 1
Latent Space · Oct 17, 2026