Nedyalko Prisadnikov

Nedyalko Prisadnikov, Danda Pani Paudel, Yuqian Fu et al.

Vision Transformers (ViTs) dominate self-supervised learning (SSL). While they have proven highly effective for large-scale pretraining, they are computationally inefficient and scale poorly with image size. Consequently, foundational models like DINO are constrained to low-resolution processing. A recent foveal-inspired transformer achieves resolution agnosticism by iteratively processing a fixed-size context of multi-zoom patches. This model demonstrated promising results via supervised learning, utilizing a sequential, recurrent-like process without backpropagation through time. To unlock its potential as a foundational backbone, we introduce a novel sequential-to-global SSL framework based on DINO's self-distillation objective. Supported by an efficient integral-image patch extraction method, our approach enables large-scale pretraining for image-size agnostic vision encoders. We achieve competitive performance on ImageNet-1K and downstream classification tasks, maintaining a constant computational budget regardless of input resolution.

Nedyalko Prisadnikov, Wouter Van Gansbeke, Danda Pani Paudel et al.

Panoptic segmentation is an important computer vision task, where the current state-of-the-art solutions require specialized components to perform well. We propose a simple generalist framework based on a deep encoder - shallow decoder architecture with per-pixel prediction. Essentially fine-tuning a massively pretrained image model with minimal additional components. Naively this method does not yield good results. We show that this is due to imbalance during training and propose a novel method for reducing it - centroid regression in the space of spectral positional embeddings. Our method achieves panoptic quality (PQ) of 55.1 on the challenging MS-COCO dataset, state-of-the-art performance among generalist methods.

Carmelo Scribano, Mohammad Mahdi, Nedyalko Prisadnikov et al.

Pretrained vision foundation models deliver strong performance across tasks with limited fine-tuning. However, their Vision Transformer (ViT) backbones impose high inference costs, limiting deployment on resource-constrained devices. In this work, we accelerate large-scale pretrained ViTs while preserving their feature extraction capabilities by exploiting the intrinsic convolution-like behavior of some attention heads. Specifically, we introduce an efficient depthwise convolution-based layer that serves as a drop-in replacement for these heads. Additionally, we propose simple strategies to identify which heads can be replaced and introduce a fine-tuning procedure that recovers downstream task performance. Across both image classification and segmentation tasks, our method achieves 17-20\% percent inference speedup with minimal performance degradation. We validate the approach through detailed derivations, extensive experiments, and efficiency benchmarks. The reference implementation is publicly available.

Nedyalko Prisadnikov, Danda Pani Paudel, Yuqian Fu et al.

This position paper argues that the next generation of vision encoders should be image size agnostic and task driven. The source of our inspiration is biological. Not a structural aspect of biological vision, but a behavioral trait -- efficiency. We focus on a couple of ways in which vision in nature is efficient, but modern vision encoders not. We -- humans and animals -- deal with vast quantities of visual data, and need to be smart where we focus our limited energy -- it depends on the task. It is our belief that vision encoders should be dynamic and the computational complexity should depend on the task at hand rather than the size of the image. We, also, provide concrete first steps towards our vision -- a proof-of-concept solution for image classification. Despite classification being not very representative for what we are trying to achieve, it shows that our approach is feasible and promising.