The paper introduces Gated Token Recurrence (GTR), a softmax‑free recurrent vision backbone that replaces global softmax attention with gated linear attention, alternating scan directions, and enhanced SwiGLU blocks. GTR is distilled from a DINOv3 teacher using only final‑layer patch‑token alignment, and achieves strong performance on COCO object detection (58.9 box AP) with very low latency (1.908 ms on an RTX 4090). The backbone also transfers to multiple dense prediction tasks and runs efficiently on edge hardware via a specialized CUDA operator and TensorRT deployment.
arXiv:2608.22368v1 Announce Type: new Abstract: While linear attention is a compelling mechanism for high-resolution object detection due to its reduced cost for global token mixing, converting the S...
arXiv:2606. 03748v1 Announce Type: cross Abstract: Real-time vision demands models that are accurate, efficient, and simple to deploy across diverse hardware.
RT-DETRv4 introduces a cost‑effective distillation framework that leverages Vision Foundation Models to improve lightweight real‑time object detectors. The method uses a Deep Semantic Injector to embed high‑level VFM representations into deep detector layers and a Gradient‑guided Adaptive Modulation strategy to adjust semantic transfer intensity based on gradient norms. These techniques yield consistent performance gains across DETR‑based models without adding inference overhead, achieving state‑of‑the‑art COCO AP scores at high FPS.
arXiv:2605.15195v2 Announce Type: replace Abstract: Recent feed-forward reconstruction models, such as VGGT, have proven competitive with traditional optimization-based reconstructors while also prov...
EdgeCrafter presents a compact Vision Transformer framework tailored for edge dense prediction tasks such as object detection, instance segmentation, and human pose estimation. By employing task-specialized distillation and an edge-friendly encoder-decoder design, the ECDet backbone achieves 51.7 box AP with under 10 M parameters, while its extensions reach 48.4 mask AP and 74.8 keypoint AP using only COCO annotations. Pretraining on Objects365 further boosts performance, demonstrating that lightweight ViTs can compete with CNN-based models on resource-constrained devices.
arXiv:2605.12491v2 Announce Type: replace Abstract: Vision Transformers (ViTs) learn rich visual-semantic representations through all-to-all self-attention among patch tokens. However, this design im...
The paper introduces reViT, a recurrent vision transformer that uses a single Transformer block repeatedly to achieve the accuracy of a full-depth encoder while keeping inference FLOPs comparable. It achieves depth‑specific transformations by expressing each block’s feed‑forward network as a convex combination of a small shared expert bank, with a continuous depth coordinate guiding the mixture. Experiments on ImageNet‑1k and DINOv2 distillation show that reViT matches or surpasses state‑of‑the‑art models, reduces stored parameters by about 70%, and supports elastic‑depth deployment from a single checkpoint.
arXiv:2607.09086v2 Announce Type: replace Abstract: We present Subtoken Vision Transformer (SubViT), a selective image tokenization method for fine-grained visual recognition. Standard Vision Transfo...
arXiv:2606. 01503v1 Announce Type: cross Abstract: Unified vision-language models (VLMs) integrate visual understanding and visual generation within a single autoregressive backbone, but their joint training is computationally expensive and largely overlooked from an efficiency perspective.
arXiv:2606. 00746v1 Announce Type: cross Abstract: Vision foundation models are bottlenecked by the quadratic cost of self-attention, which limits usable resolution and increases the cost of large-scale pretraining.
arXiv:2607. 25527v1 Announce Type: cross Abstract: Unifying visual understanding and generation in one model holds immense promise, but remains challenging and expensive due to heavy compute and data demands and conflicts between the visual features needed for these two capabilities.
ProgResViT is an input‑adaptive Vision Transformer that processes images progressively across multiple rounds, starting with a low‑resolution image and a narrow subnetwork and refining the prediction with higher resolution and a wider subnetwork if needed. The method introduces Progress‑Conditioned Soft Gating (PSG) to share a single backbone across rounds while conditioning token fusion and layer outputs on the current round, block, and input resolution. Experiments on DeiT show improved accuracy‑compute trade‑offs compared to adaptive‑width, adaptive‑depth, and dynamic‑token baselines, and the design also benefits self‑supervised DINO representations and downstream semantic segmentation.