In modern large language models (LLMs), increasing the context length is crucial for improving comprehension and coherence in long-context, multi-modal, and retrieval-augmented language generation. While many recent transformer models attempt to extend their context length over a million tokens, they remain impractical due to the quadratic time and space complexities. Although recent works on linear and sparse attention mechanisms can achieve this goal, their real-world applicability is often limited by the need to re-train from scratch and significantly worse performance. In response, we propose a novel approach, Hierarchically Pruned Attention (HiP), which reduces the time complexity of the attention mechanism to $O(T \log T)$ and the space complexity to $O(T)$, where $T$ is the sequence length. We notice a pattern in the attention scores of pretrained LLMs where tokens close together tend to have similar scores, which we call "attention locality". Based on this observation, we utilize a novel tree-search-like algorithm that estimates the top-$k$ key tokens for a given query on the fly, which is mathematically guaranteed to have better performance than random attention pruning. In addition to improving the time complexity of the attention mechanism, we further optimize GPU memory usage by implementing KV cache offloading, which stores only $O(\log T)$ tokens on the GPU while maintaining similar decoding throughput. Experiments on benchmarks show that HiP, with its training-free nature, significantly reduces both prefill and decoding latencies, as well as memory usage, while maintaining high-quality generation with minimal degradation. HiP enables pretrained LLMs to scale up to millions of tokens on commodity GPUs, potentially unlocking long-context LLM applications previously deemed infeasible.

We propose an end-to-end online multi-object tracking (MOT) framework with a systematized event-aware loss, which is designed to control possible occurrences in an online MOT situation and compel the tracker to take appropriate actions when such events occur. Training samples from real candidates using a simulation tracker are generated, and a systematized event-aware association matrix is constructed for every frame to enable the tracker to learn the ideal action in a running environment. Several experiments, including ablation studies on various public MOT benchmark datasets, are conducted. The experimental results verify that each event affecting the tracking measure can be controlled, and the proposed method presents optimal results compared with recent state-of-the-art MOT methods.

We propose Uncertainty Augmented Context Attention network (UACANet) for polyp segmentation which considers an uncertain area of the saliency map. We construct a modified version of U-Net shape network with additional encoder and decoder and compute a saliency map in each bottom-up stream prediction module and propagate to the next prediction module. In each prediction module, previously predicted saliency map is utilized to compute foreground, background and uncertain area map and we aggregate the feature map with three area maps for each representation. Then we compute the relation between each representation and each pixel in the feature map. We conduct experiments on five popular polyp segmentation benchmarks, Kvasir, CVC-ClinicDB, ETIS, CVC-ColonDB and CVC-300, and our method achieves state-of-the-art performance. Especially, we achieve 76.6% mean Dice on ETIS dataset which is 13.8% improvement compared to the previous state-of-the-art method. Source code is publicly available at https://github.com/plemeri/UACANet

This paper proposes the fusion of a saliency map and a deep feature-based correlation filter to enhance tracking accuracy by reflecting spatial attention and foreground information in the tracking process. The saliency map enables the tracker to focus on a salient object region. The target foreground region is roughly segmented from the background region using a pixel-wise likelihood map derived from the color model and the shape model. Given that only visual information in the foreground region is used, the model is prevented from learning background, and the tracker is made robust to background change. The proposed saliency map can be easily combined with any tracking methods by providing the map as the weight value to the target response map. The saliency map is combined with a correlation filter-based tracker, and we prove that the saliency map successfully improves tracking performance. Experiments are conducted to validate the proposed method on public benchmark datasets. The proposed method achieves remarkable results compared with existing state-of-the-art trackers and successfully improves the tracking performance combined with two version of correlation filter-based methods.

In this paper, we propose a salient region-based tracking method that discriminates the exact target region from background by using a probabilistic color model. The color model is updated using image pixels included in salient region. From the extracted salient region, we derive shape model which can be combined with color model that enable the tracker to be robust when the color distribution of target object is similar with other objects. Additionally, we adopt template matching weighted by the shape model to discriminate the target when the background has very similar color distribution with target object. The weight between color matching and template matching is automatically determined based on the confidence of the response map. The proposed method is robust to scale change, object transformation, and rotation. In experiments on public datasets, the proposed method achieved a higher result compared with existing state-of-the-art methods in terms of Expected Overlap Ratio (EAO) only using color model and template matching. The internal analysis proves that the combination of salient region and shape model can increase the tracking performance.