Efficient ViTs

In this work, we show that a single Transformer block, applied recurrently, can match the accuracy of a full-depth vision encoder at comparable inference FLOPs without intermediate feature distillation. reViT restores depth-specific transformations by representing the FFN at each recurrent depth as a convex combination of a small shared expert bank. A continuous normalized-depth coordinate programs this mixture, defining a resampleable trajectory through FFN parameter space. We evaluate this design in two regimes: supervised ImageNet-1k training and distillation from a DINOv2 teacher. Across both regimes, controlled adaptations identify weight-space merging as the strongest tested MoE family at a matching one-FFN budget, ahead of the token-dispatch and output-mixture alternatives. Trained from scratch, reViT-B/16 attains DeiT III accuracy with about 70% fewer stored parameters. An 8-experts model distilled using only the teacher's output features retains nearly all of its DINOv2 teacher's linear-probe accuracy and transfers across classification, segmentation, and depth prediction. Elastic-depth training allows one checkpoint (trained model) to operate at multiple tested depths by resampling the same normalized coordinate interval. For fixed-depth deployment, the recurrent block can be materialized as a conventional dense graph, removing online routing and merging without changing the one-FFN-per-depth compute but expanding deployment storage.

To adapt billion-parameter Vision Transformers for long-video understanding, recent methods freeze the backbone and train lightweight convolutional modules. While effective for parameter-efficient training, existing adapters do not reduce inference-time computation, leaving scalability with respect to video length largely unaddressed. Token selection can reduce attention cost by pruning redundant tokens, but it breaks the spatial grid structure required by convolutional adapters. This forces an expensive dense reconstruction, nullifying much of the potential speedup. We address this by introducing native sparse 2D convolutions, a primitive that allows these adapters, for the first time, to operate directly and efficiently on dynamically pruned token sets. We integrate this primitive into SV-TAD, an adapter framework for temporal action detection, reducing VideoMAEv2-L computation by up to 64% and achieving 2.2x faster inference, while maintaining state-of-the-art accuracy on THUMOS-14 and ActivityNet-1.3. When scaled to InternVideoNext-L, our approach surpasses the previous state of the art at roughly half its computational cost. Moreover, the sparse formulation naturally supports auxiliary task tokens, which improves fine-grained assembly detection on ATTACH.

Representation structure is crucial for understanding Vision Transformer (ViT) architectures and their generalization behavior. However, prior studies neither isolate nor analyze module-level features nor investigate how their interactions contribute to performance estimation. In this work, we conduct the first rigorous analysis of feature information across diverse architectural scales, empirically uncover the relationship between ViT representation and generalization behavior, and leverage these insights to guide efficient ViT design. Our contributions are fivefold: Across diverse architectural scales, 1) We identify feature collapse at initialization, which leads to redundancy, and propose a reduction scheme to mitigate this issue. 2) We quantify feature information using entropy and the minimum eigenvalue, demonstrating that these metrics serve as reliable indicators for generalization prediction. 3) We show that feature in the token space provides a more faithful representation than those in embedding space. 4) We discover an unexpected finding: features produced by linear submodules within ViT layers are critical for the prediction of generalization performance. 5) Our proposed proxy improves the correlation ranking by 18-48% over prior baselines and can effectively identify ViT architectures that achieve higher accuracy at lower or comparable computational cost.

The Visual Geometry Grounded Transformer (VGGT) marks a significant leap forward in 3D scene reconstruction, as it is the first model that directly infers all key 3D attributes (camera poses, depths, and dense geometry) jointly in one pass. However, this joint inference mechanism requires global attention layers with extremely long sequences that causes a significant latency bottleneck. In this paper, we propose blockwise clustered attention (BC attention) to accelerate the global attention layers in VGGT. By limiting the clustering within HW-friendly neighborhood blocks, BC attention reduces the computation overhead of query clustering as well as the costly data movement between on- and off-chip memory. This enables BC attention to scale to long sequences and deliver practical latency improvements on GPUs. Moreover, we introduce a hashing hyperplane calibration method and a threshold-based error compensation method to reduce clustering errors efficiently, which is a bottleneck in the current clustered attention mechanism. Overall, our experiments on GPU demonstrate that calibrated BC attention accelerates the global attention layers by 2.10-2.63$\times$ and the whole backbone by 1.77-2.35$\times$ with negligible loss (1%) for large scenes. With a small performance loss (< 5%), calibrated BC attention further achieves a 2.26-2.87$\times$ latency improvement on the global attention layers and a 1.90-2.55$\times$ improvement on the backbone.

Training-free token reduction accelerates vision transformers by removing redundant tokens across layers, recovering most of the original accuracy at a fraction of the compute. These methods, however, are designed and evaluated primarily on clean data, and under real-world distribution shift their accuracy gap to the uncompressed model widens with the removal rate. We show that this gap is governed by the reduction schedule, the depth profile of removal, usually left fixed as an implementation detail. Concretely, we introduce a one-parameter late-concentrated power-law schedule that consistently improves out-of-distribution accuracy over flat at no extra inference cost. On ImageNet-C with DeiT-S, the late schedule closes 83% of that gap at a 26% compute reduction (+1.17pp), and 99% of it at a lighter 7% reduction (+0.26pp). The gain cannot be attributed to retaining more tokens or using extra compute: held to flat's compute, the late schedule removes more tokens in total and leaves fewer tokens at the end, yet still wins. Single-layer probes point to a mechanism: earlier reductions perturb features that pass through more remaining layers, front-loading reduction error in depth. The effect is broad, holding across five token-reduction methods (ToMe, EViT, ATS, ATC, PiToMe), nine backbones, all ImageNet-C corruption types, eight further shift suites, and two further modalities, video and vision-language QA. It is also specific to shift, still positive on clean and rising monotonically to ~4x that at the highest severity 5. The schedule keeps its gain under six test-time adaptation methods, and needs no per-input or per-domain tuning.

Feed-forward 3D vision models such as VGGT have achieved remarkable progress, unifying camera estimation and dense scene reconstruction in a single pass. However, their quadratic global attention makes long image sequences expensive, while existing sparse methods may favor highly attended yet value-redundant regions. To address these limitations, we introduce VASC, a training-free sparse attention method combining value-aware block selection and execution-aware cross-layer memory. Our value-aware block selection integrates pooled query--key relevance with neighboring value contrast, reducing redundancy while preserving query-relevant and distinctive content. Cross-layer memory tracks unserved demand across layers and updates this state according to actual execution, enabling previously underserved blocks to compete under a fixed computation budget. Experiments on 7Scenes and NeuralRGB-D with VGGT and $π^3$ demonstrate improved pose estimation and reconstruction quality compared with FasterVGGT, together with up to $2.29\times$ faster inference than dense VGGT. Code is available at https://github.com/kosakayamahoo-design/VASC.

Predicting canopy height from medium-resolution satellite imagery is a common and scalable approach for assessing the condition of the world's forests, which play a crucial role in climate change mitigation. While Transformer-based architectures have shown strong performance in many domains, their straightforward application to dense (i.e., pixel-level) regression tasks often yields suboptimal results. In particular, the patch size has a crucial impact on the model performance. In this work, we consider pixel-level attention schemes and show that the resulting models generally outperform those relying on larger patch sizes. However, pixel-level attention can be a prohibitively resource-intensive operation. For this reason, we conduct an extensive experimental study using efficient attention variants to identify favorable trade-offs between prediction quality and resource requirements, facilitating the practical deployment of the proposed models. In addition, we perform a comprehensive comparison with several well-established models in the field and show that, with suitable hyperparameter choices, Transformer-based architectures can outperform competing approaches. Our findings provide practical guidance for designing models for pixel-level regression tasks on medium-resolution satellite imagery, including canopy height and biomass estimation, soil moisture mapping, and yield forecasting.

Fine-grained visual perception enables vision-language models to distinguish subtle attributes and ground their answers in visual evidence. In high-resolution scenes, processing the whole image at greater resolution spends visual tokens on irrelevant content, while isolated crops can lose the context needed to interpret the selected evidence. We introduce EviViT, a lightweight attachment that learns where a pretrained vision transformer should acquire detail. Human visual-search traces supervise a question-conditioned evidence density, which guides regional re-reading from the original pixels and the allocation of visual tokens. A sparse, coordinate-aware bridge then connects the regional features to the global scene, allowing the host to interpret precise evidence in context. Learned with the host backbone frozen, the attachment serves both the base model and compatible post-trained descendants without refitting. Experiments across nine hosts show consistent gains in average fine-grained accuracy. Matched-budget comparisons further show that EviViT outperforms global-only processing at every tested token ceiling while using fewer visual tokens.

Linear Vision Transformers (ViTs) are designed to replace the attention in Softmax ViTs with the linear-complexity attention operator for more efficient token routing, but they require from-scratch pre-training and typically underperform the original Softmax version. How to initialize linear ViTs both efficiently and effectively still remains unclear. In this work, we explicitly ask: given that most foundation ViTs are built on the mainstream Softmax attention, can linear ViTs benefit from their pre-trained weights? Recent works on Attention Transfer show that attention is the effective transferable component between Softmax ViTs, suggesting attention alone suffices for such reuse. However, we find the opposite for Softmax-to-linear transfer. The attention weights are operator-specific: copying them barely helps, and is sometimes even worse than random initialization. Instead, the attention's token routing behavior can be recovered through distillation with a proper loss design, letting linear ViTs reduce the gap and even match Softmax ones. In contrast, the MLP weights, which carry the learned representation, are operator-agnostic: they can be transferred by simple direct copying, which already carries most of the benefit of the pre-trained weights. Thus, copying MLPs can serve as an effective foundation for Softmax-to-linear transfer: paired with the distilled attention, linear ViTs eventually close the remaining gap and even surpass Softmax ones. These findings hold consistently across various linear ViT variants, different model sizes, and diverse datasets. We hope this study deepens the understanding of reusing pre-trained weights across attention operators: copy what stays the same and distill what differs, to recover the benefit across the Softmax-to-linear boundary.

3D vision transformers such as VGGT predict camera poses and scene geometry from multi-view images in a single forward pass, but their global attention over all concatenated view tokens dominates computation as the number of views grows. To reduce this cost, SparseVGGT and HeSS sparsify attention at the block level, and both retain blocks with high attention probability. However, we observe that attention probability poorly predicts how much the model's behavior actually changes when a block is removed, and we show that this mismatch is why performance collapses as sparsity increases. In this paper, we propose ReSS (ReSidual-ReStoring Sparse Attention), which recasts block selection from a problem of maximizing the retained attention mass to one of minimizing the drift that sparsification leaves in the residual stream. We introduce a drift score that quantifies how much each block shifts the residual, and, since the drift of a drop set depends on the directions of the contribution vectors rather than on their magnitudes alone, an iterative residual restoration procedure that refines the drop set as a whole. Across three backbones and five datasets, ReSS preserves dense performance better than prior methods at matched sparsity. Two further results support drift as the quantity that governs the cost of sparsification: maximizing drift degrades performance faster than random selection, and plotted against realized drift instead of sparsity, all methods fall approximately onto a single curve. Code is available at https://github.com/libary753/ReSS.

Vision Transformers (ViTs) incur quadratic self-attention cost in the number of tokens. Most token-reduction methods adapt token identities within a prescribed layer-wise compression schedule, or search a static mask offline, and thus limit online adaptation of when and how much to prune. We propose DORA (Dynamic Online Reinforcement Agent), which learns an input-adaptive pruning policy itself for frozen ViTs. At each eligible block, a hierarchical actor decides whether to prune, how many tokens to remove, and which tokens to remove from each image's evolving representation. Because early deletions change the states observed by later decisions, DORA formulates pruning as a finite-horizon Markov decision process. Complete-prefix shadow evaluations convert final-prediction fidelity into localized per-step credit, while closed-loop accuracy feedback adjusts the fidelity penalty toward a shared accuracy-drop target. A privileged critic and all shadow computations are training-only. Deployment retains the frozen backbone and a lightweight actor that applies hard deletion and packed variable-length FlashAttention, converting token reduction into measured speedups. On ImageNet-1K with DeiT-Base, DORA reduces FLOPs by 38.4% relative to the uncompressed backbone within one percentage point of accuracy loss. Averaged across four ViT-type backbones at matched accuracy, DORA uses 13.2% fewer FLOPs and achieves 32.4% higher throughput than the corresponding per-backbone baseline means. Under zero-shot transfer to ImageNet-A, these gains widen to 20.3% and 45.6%, respectively.

Comparisons in video self-supervised learning often evaluate complete training recipes rather than isolating the method itself: architecture, objective, data exposure, schedule, scale, and decoder capacity can all vary at once. This makes it hard to identify which choices yield motion-prioritized representations, whose gains concentrate on frame-to-frame change while retaining useful appearance. We address this with a matched $4 \times 6 = 24$ architecture-objective study at roughly 170M ~ 190M encoder scale on $\sim$1.7M OpenVid and Moments-in-Time v2 clips for 8 epochs, and propose TT-VidT. TT-VidT combines a DINOv3-initialized ViT-B/16 per-frame spatial path with a compact Temporal Transfer Layer, trained by Diff Compression to reconstruct target frames from a first-frame appearance anchor and frame-specific motion tokens. The sweep shows that TT3D with Diff Compression, not either component alone, enters the strongest motion-sensitive regime, and decoder ablations favor a compact video-pretrained decoder. In final comparison, TT-VidT leads Jester, Something-Something V2, ARID, and Diving48 fine-tuning simultaneously, improving over the strongest non-TT row by 54% ~ 121%, while using 48% fewer encoder FLOPs than DisMo and 55% fewer than VideoMAE or V-JEPA2. HMDB51, IARD, and EPIC-Kitchens bound the claim.

Current Transformer-based tracking methods typically stack multiple Transformer blocks with separate parameters to model interactions between the target template and the search region for target localization. These trackers often incur substantial parameter overhead from stacked blocks, making their deployment on resource-limited devices difficult. To address this, we propose a parameter-efficient Transformer tracking framework, dubbed LoopTrack, which repeatedly applies a set of Transformer blocks with shared parameters to interact features in a looped architecture for tracking, significantly reducing the number of parameters. To further exploit target cues, we present two lightweight designs, including target-aware looping (TAL) and gated target memory (GTM). The former applies intermediate target information generated by one loop to guide feature interaction in the subsequent loop, enabling progressive feature refinement, while the latter maintains a compact memory across frames, which is incorporated into the loop process to provide long-term information to the tracker, mitigating temporal drift in tracking. Compared to existing Transformer trackers, LoopTrack enables multiple rounds of feature interaction with fewer model parameters, making it resource-friendly for deployment. In extensive experiments on multiple datasets, LoopTrack shows a favorable accuracy-parameter trade-off. In particular, our LoopTrack$_{\rm One}$, with a single shared Transformer block, achieves 66.2% SUC score on LaSOT with only 3.4M parameters, while LoopTrack$_{\rm Three}$, using three shared blocks, achieves 69.3% SUC score with 6.4M parameters, surpassing existing parameter-efficient tracking methods with comparable or larger model size. With LoopTrack, we aim to establish a simple yet strong baseline for parameter-efficient Transformer tracking. Our code and models will be released.

A generalist vision encoder must capture semantic, spatial, language-aligned, and action-relevant cues within a unified representation, yet softmax attention underlying today's most capable visual backbones becomes prohibitively expensive at high resolution. A natural attempt to address both challenges is to distill multiple specialist teachers directly into an efficient architecture. We find that directly coupling these objectives degrades representation quality, as the student must simultaneously reconcile heterogeneous capabilities and adapt them to a different token-mixing architecture. We introduce IronViT, built on a simple principle: consolidate capabilities before constraining computation. IronViT first distills complementary specialists into a softmax attention capability bridge, then progressively transfers the consolidated representation to a hybrid softmax-linear attention encoder. A purpose-built data pipeline further curates the distillation corpus for higher information density and broader domain coverage. Across recognition, retrieval, dense prediction, multimodal understanding, and robotic learning, IronViT is competitive with leading specialist and generalist vision encoders. The softmax bridge achieves the strongest aggregate performance in multimodal understanding and robotic learning among the evaluated backbones, while the hybrid encoder retains broad transfer performance with an efficiency advantage that grows with input resolution. Together, these results show that consolidating capabilities before architectural conversion can yield a generalist visual encoder without inheriting the prohibitive high-resolution cost of conventional softmax attention.

Audio-Visual Semantic Segmentation (AVSS) aims to identify, segment, and classify sound-emitting objects in video frames. Previous Transformer-based AVSS approaches largely inherit design principles from image segmentation models. Recent studies show that these image segmentation models contain redundant components that contribute little to the segmentation performance. Following this insight, we propose Encoder-only Audio-Visual Segmentation (EASE). EASE runs at up to 365 FPS, 3x faster than prior State-of-the-Art (SotA) AVS models at comparable accuracy, and trains in under 11 GPU-hours. Furthermore, we achieve SotA AVSS performance across different backbones and input resolutions. Our results demonstrate that AVSS can be both simpler and faster, providing a scalable foundation for future research and real-time applications. Code, model weights, and samples are available at https://ease-avs.notion.site

Methods operating on Vision Transformer (ViT) feature spaces typically rely on Euclidean distance or cosine similarity. This assumes that every direction is equally meaningful, but there is no reason to believe the true task geometry has this property. The task-sensitive geometry of the feature space is given by the pullback metric $g(F) = J(F)^\top J(F)$, where $J$ is the Jacobian of the decoder's output fed to a task-specific distance, with respect to the features. Storing the full $g$ is infeasible at modern scales, and for dense outputs such as depth maps even forming $J$ is impractical. We show that whether a low-rank approximation of this metric can be learned depends on the model-decoder pair, and we characterize this with a matrix-free diagnostic $κ_{cap}(r)$ computable with a low number of Jacobian-vector products. For tractable pairs, we develop the Spectral Pullback Network (SPN), which learns a low-rank version of the metric from randomized power iteration, and we distill it into a $310$K-parameter importance head that predicts token importance directly from the features. When the Jacobian spectrum is too spread out for a low-rank approximation, passing the decoder's input features through a VAE bottleneck can restore tractability. Across DPT, DINOv2, CLIP, and VGGT backbones, $κ_{cap}(r)$ predicts which learned-metric architectures are viable. The importance head reaches Spearman $ρ= 0.998$ on DINOv2 CLS, and our geometric token pruning reduces the additional depth error of ToMe-based token selection by $25\%$ on DPT depth at prune ratio $0.5$, without fine-tuning the ViT. Project page: https://cyberiada.github.io/TaskInducedViTs/

Self-attention-based vision backbones perform well on dense prediction, but the quadratic computational cost of global softmax attention limits their efficiency as image resolution increases. We introduce Gated Token Recurrence (GTR), a softmax-free recurrent vision backbone that combines gated linear attention, alternating spatial scan directions, and spatially enhanced SwiGLU blocks. GTR is distilled from a detection-specialized DINOv3 teacher using only final-layer patch-token alignment through a linear projection and squared $\ell_2$ loss, without masked-token prediction or intermediate-layer supervision. With Objects365 detector pre-training, GTR-L achieves 58.9 box AP on COCO \texttt{val2017} with 1.908,ms median batch-one latency under compiled FP16 execution on an RTX4090. The same backbone also transfers to instance segmentation, pose estimation, oriented detection, semantic segmentation, and monocular depth estimation. In an isolated kernel benchmark, our specialized chunkwise CUDA operator is $4.0\times$ faster than FLA v0.5.0 at 1.6K tokens on RTX4090. TensorRT deployment on DRIVE AGX Thor achieves 2.282--8.769,ms median batch-one latency across the evaluated models. These results show that recurrent token mixing can provide an efficient alternative to global softmax attention for high-resolution dense prediction and edge deployment. Project page: https://intellindust-ai-lab.github.io/projects/GTR/

While Mamba-based models have shown strong potential for long sequence modeling, adapting them to vision is challenging due to the requirement of local neighborhood correlations and multi-directional spatial contexts for visual understanding. In this paper, we present LiAuto-MindViT, a novel hybrid vision backbone that synergizes the strengths of CNNs, Mamba, and Transformers. The core of our design is the Adaptive Bidirectional Mamba (ABM), which eliminates the directional bias of unidirectional SSMs through bidirectional selective scanning with learnable alpha blending, enabling content-adaptive directional fusion without the overhead of exhaustive multi-path routing. To further accelerate inference, we propose a deployment-friendly Reparameterized ConvSE (RepConvSE) module that leverages structural reparameterization to reduce latency and memory access overhead. Extensive experiments demonstrate that LiAuto-MindViT achieves state-of-the-art performance on image classification, object detection, and semantic segmentation while enabling efficient inference through reparameterization.

Feed-forward visual geometry models such as the Visual Geometry Grounded Transformer (VGGT) have recently enabled direct 3D reconstruction from multi-view images. Despite their promising performance, these models scale quadratically with the number of input views due to their global attention mechanism, resulting in substantial latency for long sequence inputs. There have been some recent efforts to accelerate VGGT, but they primarily focus on reducing \emph{token redundancy} through token merging or key/value sparsification. Our work resolves this bottleneck from a different perspective by investigating \emph{architectural redundancy} in visual geometry transformers. We show that the multi-head attention modules in VGGT's global-attention layers contain substantial architectural redundancy, with only a subset of heads carrying critical geometric information. In light of this observation, we propose VGGT-Prime, a compute-adaptive mixture-of-heads model that resolves this redundancy to accelerate visual geometry transformers while maintaining competitive reconstruction quality. The key idea of VGGT-Prime is to estimate the appropriate computation level for each global-attention head using a lightweight router and then dynamically assign each head to different computation modes. Extensive experiments on multiple datasets demonstrate that VGGT-Prime can achieve an {$8\times$} inference speedup over VGGT while maintaining competitive performance on camera pose, depth, and point-cloud predictions. We further show that VGGT-Prime is complementary to existing acceleration methods, such as token merging, further improving inference speed by up to $14{\times}$ over VGGT. An overview of our work is available on our project page.

Vision Transformers devote most of their parameters to MLPs for channel mixing, but still rely on quadratic multi-head self-attention for token interactions. While linear attention fixes the complexity problem, bringing it down to O(N), it is usually just paired with the same fixed-activation MLP as before. Kolmogorov-Arnold Networks take a different approach, placing learnable univariate functions on the edges instead. However, existing vision KANs either retain standard attention or remove attention entirely, so the two ideas have not been effectively combined. We introduce LKAT (Linear Kolmogorov-Arnold Transformer) to close this gap: an isotropic ViT-style encoder that couples chunk-wise Gated Linear Attention with a two-layer KAN feed-forward block, backed by an I/O-aware fused RBF-KAN kernel to make radial-basis grid functions efficient in practice. Under a shared DeiT-style training recipe, LKAT-B outperforms ViT-B/16, ViT-5-B, and Mixer-B/16 on ImageNet-100, while Tiny, Small, and Base variants scale consistently on CIFAR-10/100. ImageNet-100 pretraining also transfers effectively to CIFAR fine-tuning, suggesting that gated linear attention and KAN-based radial basis functions provide complementary inductive biases for mid-scale visual representation learning. Code: https://github.com/mehizelali/linear-kan-transformer

Recent findings indicate that Vision Transformers settle into locally similar computational phases, implying a level of depthwise computational redundancy. However, existing methods to exploit this redundancy either fail to reduce inference compute or severely degrade model expressivity. In this work, we formalise a unified view of block redundancy that decouples the geometry from specific surrogate interventions. We then introduce Transformer-Within-Transformer (TWT), a post-hoc method that fuses contiguous groups of redundant layers into a single learned surrogate layer. TWT reduces parameter count and inference compute while remaining competitive with original models using half the depth on natural images, and in several downstream histopathology settings, TWT matches or even improves on the original baseline.

Automated classification of sewer defects is essential for infrastructure condition assessment and maintenance decision-making, but existing deep learning methods struggle to balance classification accuracy and computational complexity in large-scale multi-label scenarios. This study develops Sewer-Transformer-ML, a hierarchical vision Transformer with multi-level feature fusion, together with two lightweight architectures, Sewer-MobileNet-ML and Sewer-Mobile-TransNet, for resource-constrained inspection scenarios. On the Sewer-ML test set, Sewer-Transformer-ML-Base achieved an $F2_{\text{CIW}}$ of 65.68% and an $F1_{\text{Normal}}$ of 92.68%, ranking first on the public leaderboard and exceeding the second-ranked method by 7.6 percentage points in $F2_{\text{CIW}}$. Sewer-MobileNet-ML achieved an $F2_{\text{CIW}}$ of 65.73% with only 17 M parameters, representing an approximately 95% parameter reduction relative to the base model. Under the standard Sewer-Capsule data split, Sewer-Mobile-TransNet achieved 96.43% classification accuracy. When the training set was reduced to 1,177 images, pretraining on Sewer-ML consistently improved model performance. Ablation experiments further showed that direct concatenation was more effective for Transformer features, whereas attention-based fusion better supported multiscale CNN features. These findings provide a computational basis for automated sewer inspection, lightweight model design, and adaptation across civil infrastructure inspection platforms.

Optical flow methods typically rely on task-specific inductive biases, such as correlation volumes, feature warping, and iterative refinement, among others, to reach high accuracy. While effective, such biases constrain the model to predefined heuristics, which can limit its expressivity and lead to more complex pipelines and additional computational cost. We present FreeFlow, a hierarchical transformer built without any flow-specific components, using instead a single feed-forward encoder--decoder. FreeFlow combines three attention variants: window attention for local processing, shifted-window attention for cross-window information exchange, and a global attention operating at a reduced resolution. The resulting architecture scales naturally with model capacity, enabling a consistent accuracy gain from small to large variants. Despite the absence of standard inductive biases, FreeFlow achieves state-of-the-art results on major benchmarks, including Sintel (0.68/1.48 EPE on Clean/Final), KITTI-2015 (3.23 Fl-all), and Spring (3.192 1px), while remaining memory efficient at 1080p inference.

Semantic segmentation for autonomous driving requires reliable detection of vulnerable road users (VRUs) despite heavy class imbalance. We introduce CLFTv2, a hierarchical camera-LiDAR fusion framework replacing global ViT attention with a Swin-based multi-scale encoder and a lightweight FPN-style residual decoder. Operating in the 2D perspective domain, CLFTv2 integrates multi-scale geometric cues through shifted-window attention and per-scale residual fusion, avoiding the computational overhead of query-matching decoders. Across three driving datasets, CLFTv2 consistently improves VRU recall. On ZOD, CLFTv2-Large achieves 53.5% mIoU, improving pedestrian IoU from 35.5% to 44.9% over the prior CLFT model. On Waymo, CLFTv2 reaches 61.7% mIoU. Additionally, a modality-isolation study suggests ViT's global receptive field yields stronger fusion gains only under dense LiDAR returns. Compared to a Swin-based Mask2Former adaptation, CLFTv2 requires 1.4$\times$ fewer GFLOPs and delivers 2.2$\times$ higher throughput, while achieving comparable overall accuracy. These results demonstrate that hierarchical local-attention fusion offers an efficient, scalable alternative to global-attention and query-based decoders for real-time on-vehicle perception in intelligent transportation systems. Source code is publicly available.

Model compression is key to mitigate deployment challenges of ever growing machine learning models. In this area of research, singular value decomposition (SVD)-based compression offers a compelling trade-off between computational efficiency and model accuracy. Fisher-weighted SVD in particular provides principled, loss-aware compression. However, we find that improving the fidelity of Fisher approximation used in the compression is poorly predictive of post-compression accuracy for Vision Transformers (ViTs). Motivated by this observation, we propose FACTS, a structured Fisher Approximation tailored to Compressing ViTs with Fisher-weighted SVD, which enforces token-local aggregation while preserving within-token activation-gradient dependence. Additionally, we introduce a fast Constrained Rank Search (CoRS), that optimizes layer-wise rank allocation while adhering to a fixed floating point operation (FLOP) constraint. Extensive experiments across ViTs and hybrid architectures demonstrate that FACTS consistently improves accuracy-efficiency trade-offs without requiring finetuning. Notably, it outperforms the strongest SVD baseline by up to +5.8 percentage points (p.p.) Top-1 on Swin-B, with further gains driven by our search method. Code is available at https://github.com/MoritzTho/FACTS.

Vision Transformers (ViTs) typically process every image using a fixed input resolution and model width, even though many images can be classified with substantially less computation. We introduce ProgResViT, an input-adaptive ViT that performs inference progressively across multiple rounds. The first round processes a low-resolution image with a narrow subnetwork. Inference terminates when the prediction is sufficiently confident; otherwise, the model reuses the representations produced in the current round and proceeds with a higher-resolution input and a wider subnetwork to refine its prediction. As all rounds share a single backbone, we propose Progress-Conditioned Soft Gating (PSG), which conditions token fusion and layer outputs on the current round, block, and input resolution. On image classification, applying ProgResViT to DeiT yields better accuracy-compute trade-offs than adaptive-width, adaptive-depth, and dynamic-token baselines. With knowledge distillation, a DeiT-based ProgResViT achieves 84.9% top-1 accuracy, slightly exceeding the reported DeiT-III-S accuracy under a comparable evaluation setting. We show that the same design also provides favorable accuracy-compute trade-offs for self-supervised DINO representations and downstream semantic segmentation. Code is available at https://github.com/ds-kiel/ProgResViT.

Modern generative models, such as GANs, diffusion architectures, and autoregressive systems, now produce facial images that are nearly indistinguishable from authentic photographs. This capability makes detecting forged images increasingly difficult, raising serious concerns about identity theft, fraud, and misinformation campaigns. Our research focuses specifically on GAN-generated synthetic faces, which underpin many face-centric deepfakes, and investigates efficient detection approaches using image analysis alone. Existing detection systems rely heavily on either convolutional neural networks (CNNs) or global vision transformers. While CNNs excel at identifying texture-based local features, they struggle with broader contextual understanding. Traditional Vision Transformer (ViT) models can capture long-range structures effectively, but demand substantial computational resources. Our work explores Swin-Transformer-based architectures across three implementations: a compact Swin Transformer trained from the ground up, ImageNet-1K pre-trained Swin-Tiny and Swin-Small models adapted for binary classification, and a novel hybrid combining EfficientNet-B0's convolutional processing with a Swin Transformer backend. We evaluated all models using the 140K Real and Fake Faces dataset, which includes StyleGAN-generated fake faces alongside authentic images from Flickr and DFDC, with balanced splits for training, validation, and testing. The EfficientNetB0+Swin hybrid achieved 99% accuracy and a 99.44% recall on 5,000 test images, outperforming both pure Swin variants and a previous CNN-only baseline on this dataset. Our results suggest that combining hierarchical CNN features with shifted-window self-attention provides an efficient and computationally lightweight method for detecting GAN-generated synthetic faces.

Convolutional Neural Networks (CNNs) capture local features efficiently but struggle with global context due to their limited receptive field. On the other hand, transformers effectively capture global dependencies through self-attention but suffer from high redundancy and computational costs. Thus, to leverage the advantages of both CNNs and transformers, we propose a unified model (UniCon-Former) that aims to provide robust and efficient performance on dynamic hand gesture recognition. The unified approach helps the model to learn both local and global features. At the beginning of each transformer stage, the convolution projections help in decreasing the dimension of the input vectors of the transformer block. This creates a pyramidal structure at each transformer stage. These features enable the UniCon-Former to reduce resource usage than vanilla transformers, making it flexible for learning multi-scale and high-resolution features, which is required in hand gesture recognition. We have performed experiments with NVGesture and Briareo datasets and achieved state-of-the-art results with fewer parameters and MACs.

Vision Transformers (ViTs) demonstrate exceptional performance in computer vision but suffer from large parameter counts and quadratic computational complexity, severely limiting their deployment on resource-constrained edge hardware. While recursive weight-sharing reduces parameter counts and token merging mitigates computational and memory bottlenecks, integrating these two paradigms without costly retraining is non-trivial, leaving this intersection largely unexplored. We propose MergeOver, a post-training approach that integrates Token Merging (ToMe) into the recursively weight-shared Sliced Recursive Transformer (SReT). Through an Unmerge tracking stack, constraint-safe merge-rate adjustment, and synchronised token-mass tracking across spatial permutations, MergeOver resolves the spatial and merging constraints of this integration. We further employ a stage-wise single-shot schedule that performs token reduction at the first block of each stage and maintains a fixed sequence length throughout its subsequent recursive iterations. Benchmarked on ImageNet-1K, our selected configuration reduces top-1 accuracy by 1.47 percentage points. On the GPU, it reduces peak activation memory by 37.3% and 38.4% at batch sizes 1 and 16, while throughput decreases by 21.7% at batch size 1 but increases by 21.7% at batch size 16. On a Raspberry Pi 5 (ARM CPU), it reduces latency by 2.4% and 17.6% at batch sizes 1 and 16. These results show that MergeOver can recover a meaningful part of the throughput and memory cost that recursive weight-sharing introduces, without retraining, and provides a baseline for combining token merging with hierarchical recursive transformers.

Token-pruning policies are usually designed for a single recognition pipeline, but pretrained Vision Transformers are reused across tasks with different spatial demands. We ask which parts of a pruning policy transfer across image classification, semantic segmentation, and object detection. For each pipeline, controlled probes freeze the no-pruning checkpoint and apply a series of parameter-free reduction criteria at one eligible layer at a time without retraining. The probes reveal three differences: segmentation and detection rank the criteria differently, classification is especially sensitive to attention-based pruning in the earliest layers, and the dense tasks prefer opposite recovery endpoints. These findings motivate Task-Adaptive Pruning (TAP). Existing register tokens serve as task-agnostic storage for feature artifacts. TAP instead introduces one task register per task and activates only the current one. Its evolving state ranks tokens, distributes an exact removal budget over depth, and sets the recovery scale for dense features. At a final keep rate of $ρ=0.5$, our jointly adapted model, TAP-J, reaches $47.0$ mIoU at $1.30\times$ encoder throughput on ADE20K and $53.7$ box AP at $1.32\times$ encoder throughput on COCO while remaining competitive on ImageNet-1K.