ADE20K
PulseAugur coverage of ADE20K — every cluster mentioning ADE20K across labs, papers, and developer communities, ranked by signal.
5 day(s) with sentiment data
-
G2TM method enhances Vision Transformer efficiency across diverse architectures
Researchers have conducted a systematic study on Graph-Guided Token Merging (G2TM), a method designed to improve the efficiency of Vision Transformers (ViTs) by reducing the quadratic complexity associated with the self…
-
New CTOAC method improves low-bit quantization for Visual State Space Duality models
Researchers have developed a new post-training quantization (PTQ) method called Channel-wise Token-balanced Output-Aware Clipping (CTOAC) to address the low-bit quantization challenges in Visual State Space Duality (VSS…
-
TailProp introduces adaptive propagation for vision models
Researchers have introduced TailProp, a novel hierarchical vision backbone that utilizes a Tail Propagation Operator (TPO). TPO combines Gaussian and Cauchy stable-process propagators, allowing for adaptive spatial infl…
-
LookThere! Sparse Vision by Reinforced Selection framework reduces computation
Researchers have developed LookThere, a novel framework that uses reinforcement learning to enable vision transformers to process only the most relevant image tokens. This approach significantly reduces computational lo…
-
New SELECT method tackles catastrophic forgetting in semantic segmentation
Researchers have introduced SELECT, a novel method for Class-Incremental Semantic Segmentation (CISS) designed to mitigate catastrophic forgetting and background shift. The approach grounds new class learning in semanti…
-
GramLoop framework boosts DINOv3 dense-prediction models under distribution shift
Researchers have developed GramLoop, a novel framework designed to enhance the performance of frozen DINOv3 dense-prediction models when faced with distribution shifts. This method integrates additional computation with…
-
New source codecs enhance distributed semantic segmentation at low bitrates
Researchers have developed two novel source codecs to improve rate-distortion performance in distributed deep neural networks for semantic segmentation, particularly at extremely low bitrates. These codecs enable effici…
-
Iwin Transformer tackles ViT complexity with interleaved windows and convolution
Researchers have introduced the Iwin Transformer, a novel hierarchical vision transformer designed to overcome limitations in existing Vision Transformers (ViTs). This new architecture combines interleaved window attent…
-
New CM-GLasso framework learns interpretable visual-linguistic dependency graphs
Researchers have developed CM-GLasso, a novel framework for learning interpretable conditional-dependence structures from multimodal visual-linguistic data. This approach integrates vision-language representation learni…
-
New Neural Prior Estimator Learns Class Priors from Latent Representations
Researchers have developed a novel method called the Neural Prior Estimator (NPE) to learn class priors directly from a network's latent representations, rather than relying on explicit class counts. This approach invol…
-
New research explores uncertainty quantification and lightweight models for semantic segmentation
Researchers are exploring methods to improve the reliability and robustness of semantic segmentation models, particularly for safety-critical applications. One paper investigates the integration of uncertainty quantific…
-
New CW-BASS v2 method improves semi-supervised segmentation with foundation models
Researchers have developed CW-BASS v2, a new method for selecting pseudo-labels in semi-supervised semantic segmentation. This approach is designed to work effectively with strong, self-supervised foundation model teach…
-
New methods enhance Spiking Transformer performance on image tasks · 2 sources tracked
Researchers have introduced Spiking Local Interaction (SLI) and Adaptive Complementary Fusion (ACF) to enhance Spiking Transformers. These methods address limitations in standard Spiking Self-Attention (SSA) by introduc…
-
New SCLA-BCP method enhances Spiking Transformer attention locality
Researchers have developed a new method called Spatially Contiguous Local Attention with Boundary Continuity Pathway (SCLA-BCP) to improve the spatial locality of Spiking Transformers. This approach addresses the challe…
-
New frameworks enhance mask transformers and adapt State Space Models for missing data
Researchers have developed iFAN, a training framework designed to enhance mask transformers by aligning query ranking with mask quality and improving intermediate prediction distillation. This method addresses mismatche…
-
New methods emerge for efficient visual token pruning in AI models · 6 sources tracked
Researchers are developing new methods to optimize Vision Transformers (ViTs) and Multimodal Large Language Models (MLLMs) by pruning visual tokens, which are computationally expensive. Several papers propose novel tech…
-
DSeq-JEPA architecture enhances visual representation learning with sequential prediction
Researchers have introduced DSeq-JEPA, a novel architecture for self-supervised visual representation learning. This model builds upon the Image-based Joint-Embedding Predictive Architecture (I-JEPA) by incorporating a …
-
StaticSegFormer boosts semantic segmentation efficiency without performance loss
Researchers have developed StaticSegFormer, a novel static structured pruning method designed to enhance the efficiency of deep neural networks for semantic segmentation tasks. This method specifically targets attention…
-
New iFAN framework boosts image segmentation accuracy for mask transformers
Researchers have developed a new training framework called Inference-Aware Learning (iFAN) designed to improve the performance of plain mask transformers used in image segmentation. iFAN addresses two key issues: the mi…
-
New Bootleg method enhances self-supervised learning for AI models
Researchers have developed a new self-supervised learning method called Bootleg, which aims to combine the stability of generative approaches with the efficiency of predictive methods. Bootleg trains a model to predict …