Tsinghua AI

Canvas360 not only redefines panoramic generation with geometry-aware techniques but also delivers a dataset of 1 million samples that transforms how we approach in-context tasks.

State-of-the-art performance in endoscopic referring segmentation is now achievable through a novel attribute retrieval approach, transforming how we interpret complex medical imagery.

Traditional generative models struggle with subtle neurodegenerative changes, but Latent Drift captures clinically relevant progression by focusing on compressed semantic representations.

Annotation noise in vascular CT scans can be detected with a novel method that reveals systematic biases, improving training robustness dramatically.

Existing text-to-image models struggle to capture individual aesthetic preferences, but PIPBench reveals critical gaps in their performance that could redefine personalized image generation.

GlaKG achieves near-perfect classification while providing a transparent reasoning framework that links biomarker evidence to clinical rules, addressing the black-box nature of traditional deep learning models in healthcare.

Surpassing human performance in gaze estimation, PaGE closes the human-AI gap by over 60% while remaining lightweight for real-world applications.

ProCon achieves unprecedented anomaly detection accuracy without the need for training or pseudo-anomaly supervision, redefining the capabilities of memory-based methods.

Current avatar systems are more diverse than ever, yet foundational prior learning is often overlooked in discussions of photorealistic digital humans.

Reducing sampling steps from 50 to just 8 without sacrificing quality could revolutionize how we approach generative modeling.

Attention mechanisms can drastically improve pose sensing accuracy in the face of challenging visual conditions like occlusion and weak textures.

Unconstrained egocentric video generation now achieves unprecedented fidelity and control by disentangling hand and camera motion with a novel 3D-aware representation.

AutoMIA can generate complex 3D illusions in under 80 seconds, revolutionizing the intersection of art and computational design.

MiShield outperforms leading moderation tools by effectively identifying harmful semantics in multi-image content that appear benign in isolation.

MG-RWKV achieves state-of-the-art TFL performance with a groundbreaking O(T) complexity, redefining efficiency in audio-visual content authenticity verification.

Positional leakage in 3D masked autoencoders can be mitigated, leading to significantly improved semantic representation quality.

Transforming image quality assessment from a single score to a nuanced diagnosis of multiple quality issues could revolutionize smartphone ISP tuning.

OVOW achieves unprecedented accuracy and speed in reconstructing 4D scenes from a single video, making it a game-changer for physics simulation in AI.

Achieving a 22.3% word error rate with just 240 ms latency, LipsFlow redefines the capabilities of Visual Speech Recognition in challenging multi-speaker environments.

Current T2I models fall short in scientific reasoning, but fine-tuning on the new SciIR dataset boosts performance by over 20%.

Urban facade reconstruction can achieve superior geometric accuracy by integrating lightweight structural supervision, overcoming common pitfalls of traditional methods.

GeoEdit achieves unprecedented geometric accuracy and identity fidelity in object editing, overcoming the limitations of existing diffusion-based methods.

Flow Splatting achieves superior image quality and faster rendering speeds by efficiently modeling dynamic scenes with 4D Gaussian representations.

Effective keyframe extraction can boost MLLM performance by over 2% on complex video-guided tasks, revealing a critical link between video understanding and procedural learning.

HarmVideoBench reveals that existing benchmarks miss critical layers of harmful video understanding, while a new method boosts model accuracy by over 20%.

Teacher-forcing consistency models can accelerate autoregressive video generation by ten times, revolutionizing the training landscape for streaming applications.

Internal biological constraints can dramatically reduce errors in hand pose estimation, enabling robust tracking in metric space.

Achieving a 14-point boost in grounding accuracy, VistaRef redefines how we approach spatial orientation in AR and human-robot interaction.

MambaRaw achieves a remarkable 1.4 dB increase in PSNR at low metadata bitrates while slashing coding latency by nearly 9%, setting a new benchmark in raw image reconstruction.

REDI-Match not only sets a new benchmark in dense feature matching but also accelerates inference speed, achieving 41 FPS on a single GPU.

Training on SignNet-1M boosts sign language model robustness by improving generalization across diverse real-world conditions without sacrificing performance.

PointVG-R achieves a groundbreaking 15.86-point boost in mIoU by integrating geometric reasoning into visual grounding tasks, reshaping our approach to spatial interpretation in models.

Current vision-language models falter in streaming interaction understanding, with alarming mis-calibration leading to confidently incorrect predictions.

Embedding geometric intelligence into segmentation models can dramatically enhance the recovery of small vascular structures and improve overall topology fidelity.

PreciseDoc achieves unprecedented precision in grounding critical document elements, transforming how LMMs can interpret complex text-rich environments.

PoinTriE achieves state-of-the-art performance in point cloud video tasks while slashing memory requirements and annotation costs.

Reinforcement learning enables video-LLMs to re-watch and refine answers without the costly overhead of chain-of-thought training, achieving better performance with less computation.

DiT-Reward not only outperforms existing models in image evaluation but also accelerates inference by 1.65x without sacrificing quality.

Transforming Poisson noise into Gaussian noise can boost image denoising performance by up to 0.75 dB, even in challenging conditions.

Achieving up to 1.90X speedup in video generation without sacrificing fidelity, ScalingAttention redefines efficiency in Diffusion Transformers.

UnityShots achieves unprecedented cross-shot coherence in multi-shot video generation, outperforming open-source benchmarks and rivaling top closed-source systems.

Top algorithms in the HECKTOR 2025 challenge achieved impressive segmentation and survival prediction metrics, showcasing the power of multimodal imaging in oncology.

Achieving over 555 FPS in tactile simulations, TaCauchy delivers unprecedented accuracy in mechanical stress computation for robotics applications.

I2V models not only excel at dynamic editing but also provide a unique lens for diagnosing errors in Human-Object Interaction tasks.

GSPan redefines pansharpening by leveraging continuous Gaussian representations, enabling high-quality image fusion across arbitrary scales without retraining.

LazyMCoT achieves training-free visual grounding that rivals traditional methods while cutting inference time, redefining efficiency in multimodal reasoning.

R2RDreamer achieves spatial generalization improvements in manipulation tasks by leveraging 3D-aware data augmentation without the pitfalls of complex scene setups or sim-to-real gaps.

HUG achieves a remarkable 34% improvement over existing grasping methods by harnessing a million human grasps to empower robots with human-like dexterity.

TetherCache slashes quality drift in long-form video generation from 7.84 to 1.33, ensuring stability and coherence over extended sequences.

OmniDirector achieves unprecedented control over camera motion in video generation, enabling director-level precision without relying on scarce cross-paired data.