PRISM leverages multi-view videos and a 3D knowledge ontology to enhance embodied VLM's spatial, physical, and action understanding in retail environments.

Amirreza Rouhi, Parikshit Sakurikar, Satya Sai Reddy et al.

Multi-Layer Memory Framework enhances LLMs' long-term context retention, achieving 46.85% success rate and 56.90% six-period retention.

Sunil Tiwari, Payal Fofadiya

LatentPilot internalizes future visual dynamics via privileged supervision, achieving SOTA in VLN benchmarks with 66.3% success rate.

Haihong Hao, Lei Chen, Mingfei Han et al.

SparseDriveV2 leverages dense static vocabularies with trajectory decomposition, achieving 92.0 PDMS, challenging the necessity of dynamic proposals.

Wenchao Sun, Xuewu Lin, Keyu Chen et al.

MEDiC unifies CLIP distillation and pixel reconstruction, reaching 73.9% kNN and 85.1% fine-tuning accuracy on ImageNet-1K.

Konstantinos Georgiou, Maofeng Tang, Hairong Qi

FGOS-Net achieves 91.3% mIoU and 97.1% clDice via frequency-geometric disentanglement.

Jin Bai, Huiyao Zhang, Qi Wen et al.

GEMS framework enhances multimodal generation with memory and skills, enabling Z-Image-Turbo to surpass Nano Banana 2 on GenEval2.

Zefeng He, Siyuan Huang, Xiaoye Qu et al.

Proposed a Cross-Scale Decoder for off-road semantic segmentation, achieving 89.97 mIoU.

Seongkyu Choi Jhonghyun An

RINO achieves rotation-invariant non-rigid matching via RINONet, significantly improving 3D shape correspondence accuracy.

Maolin Gao, Shao Jie Hu-Chen, Congyue Deng et al.

Syn4Seg framework enhances GFSS performance by generating diverse synthetic images and pseudo-labels, achieving significant improvements on PASCAL-5i and COCO-20i.

Guohuan Xie, Xin He, Dingying Fan et al.

SJD-VP enhances autoregressive image generation by predicting verification to improve acceleration and quality.

Bingqi Shan, Baoquan Zhang, Xiaochen Qi et al.

DiReCT improves VideoPhy physical commonsense score by 16.7% without increasing training time.

Abolfazl Meyarian, Amin Karimi Monsefi, Rajiv Ramnath et al.

Fus3D leverages pretrained multi-view geometry transformer latents to directly regress dense SDF in under 3 seconds, enabling fast, pose-free 3D reconstruction.

Laura Fink, Linus Franke, George Kopanas et al.

GDPO-Listener generates expressive head motions via Auto-Regressive Flow Matching and Group reward-Decoupled Policy Optimization.

Zhangyu Jin, Maksim Siniukov, Deuksin Kwon et al.

AVControl leverages LTX-2 with LoRA adapters for efficient multi-modal audio-visual control, achieving high performance with minimal training steps.

Matan Ben-Yosef, Tavi Halperin, Naomi Ken Korem et al.

Latent-WAM achieves efficient end-to-end autonomous driving with spatially-aware and dynamics-informed latent world representations, scoring 89.3 on NAVSIM v2.

Linbo Wang, Yupeng Zheng, Qiang Chen et al.

EndoVGGT enhances surgical 3D reconstruction with DeGAT, improving PSNR by 24.6% and SSIM by 9.1%.

Falong Fan, Yi Xie, Arnis Lektauers et al.

VFIG uses vision-language models for complex figure-to-SVG conversion, achieving a VLM-Judge score of 0.829.

Qijia He, Xunmei Liu, Hammaad Memon et al.

Proposed a simplified, stable flow-matching generative model, scaled twice as large, with twofold FID reduction and improved unseen molecule simulation using molecular embeddings.

Charles Jones, Emmanuel Noutahi, Jason Hartford et al.

MedObvious exposes the Medical Moravec's Paradox in VLMs via a 1,880-task benchmark for clinical triage.

Ufaq Khan, Umair Nawaz, L D M S S Teja et al.