Is this you? Claim this profile to correct it, add a bio and choose the work people see first.
Claim this profileWorks58 from public data
- Open-Sora 2.0: Training a Commercial-Level Video Generation Model in $200k176
This report presents Open-Sora 2.0, a commercial-level video generation model trained for only $200k, and demonstrates that the cost of training a top-performing video generation model is highly controllable.
- 131
- 104
- KeypointNet: A Large-Scale 3D Keypoint Dataset Aggregated From Numerous Human Annotations95
This work presents KeypointNet: the first large-scale and diverse 3D keypoint dataset that contains 83,231 keypoints and 8,329 3D models from 16 object categories, by leveraging numerous human annotations.
- Classification of ASD based on fMRI data with deep learning88
Experimental results indicate that the DFS can effectively select critical FC features for classification according to the weights of input FC features and the performance of GCN classifier can be improved dramatically.
- Pointwise Rotation-Invariant Network with Adaptive Sampling and 3D Spherical Voxel Convolution75
A brand new point-set learning framework PRIN is proposed, namely, Pointwise Rotation-Invariant Network, focusing on rotation-invariant feature extraction in point clouds analysis, which shows that, on the dataset with randomly rotated point clouds, PRIN demonstrates better performance than state-of-the-art methods without any data augmentation.
- CPPF: Towards Robust Category-Level 9D Pose Estimation in the Wild54
This paper designs a novel Category-level PPF (CPPF) voting method to achieve accurate, robust and generalizable 9D pose estimation in the wild, and shows that the method is robust to noise and gives promising results under extremely challenging scenarios.
- Decoding human brain activity with deep learning54
A novel method of decoding brain activity evoked by visual stimuli by using an image-EEG-image transformation and shows that the proposed LSTM-CNN algorithm that discriminates the object classes by using EEG can be more accurate than the existing methods.
- 49
- Img2CAD: Reverse Engineering 3D CAD Models from Images through VLM-Assisted Conditional Factorization46
A novel approach that conditionally factorizes the reverse engineering of 3D CAD models from images into two sub-problems is introduced, and a TrAssembler that conditioned on the discrete structure with semantics predicts the continuous attribute values is proposed.
- Skeleton Merger: an Unsupervised Aligned Keypoint Detector45
It is demonstrated that Skeleton Merger is capable of detecting semantically-rich salient keypoints with good alignment, and shows comparable performance to supervised methods on the KeypointNet dataset, and it is also shown that the detector is robust to noise and subsampling.
- Classification of ADHD with fMRI data and multi-objective optimization34
The proposed multi-objective classification scheme addresses the imbalanced dataset problem by using a three objective SVM model with the positive and negative empirical errors being handled explicitly and separately, and effectively addresses dataset imbalanced problem from algorithm level.
- UKPGAN: A General Self-Supervised Keypoint Detector30
This work proposes UKPGAN, a general self-supervised 3D keypoint detector where keypoints are detected so that they could reconstruct the original object shape, and two modules: GAN-based keypoint sparsity control and salient information distillation modules are proposed to locate those important keypoints.
- 23
- PRIN: Pointwise Rotation-Invariant Network.23
A new point sets learning framework named Pointwise Rotation-Invariant Network (PRIN), focusing on the rotation problem in point clouds, which shows similar performance on par or better than state-of-the-art methods on part segmentation without data augmentation.
- 23
- 20
- Point cloud classification with deep normalized Reeb graph convolution19
This paper addresses the problem of classifying point clouds with unknown orientations by extracting the Reeb graph from point clouds, and then filter the graph with deep graph convolution, and shows that this method gives a competitive performance to the state-of-the-art methods and outperforms previous methods by a large margin.
- Combinatorial Q-Learning for Dou Di Zhu18
This paper studies a special class of Asian popular card games called Dou Di Zhu, in which two adversarial groups of agents must consider numerous card combinations at each time step, leading to huge number of actions, and proposes a novel method to handle combinatorial actions, which is called combinatorially Q-learning (CQL).
- Canonical Voting: Towards Robust Oriented Bounding Box Detection in 3D Scenes15
This work disentangle the direct offset into Local Canonical Coordinates (LCC), box scales and box orientations, which achieves state-of-the-art performance on three standard real-world benchmarks: ScanNet, SceneNN and SUN RGB-D.
- Robotic picking in dense clutter via domain invariant learning from synthetic dense cluttered rendering14
A novel domain-invariant Suction Quality Neural Network (diSQNN) is proposed to fuse realistic color feature and adversarial depth feature with a domain discriminator on depth extractor to reduce simulation-to-reality gap from synthetic images to low-quality RGB-D camera.
- CPPF++: Uncertainty-Aware Sim2Real Object Pose Estimation by Vote Aggregation12
This paper introduces a novel method, CPPF++, designed for sim-to-real category-level pose estimation, which significantly surpasses previous sim-to-real approaches and achieves comparable or superior performance on novel datasets.
- PACE: A Large-Scale Dataset with Pose Annotations in Cluttered Environments12
This work introduces PACE (Pose Annotations in Cluttered Environments), a large-scale benchmark designed to advance the development and evaluation of pose estimation methods in cluttered scenarios, and develops an innovative annotation system with a calibrated 3-camera setup.
- The Snowflake Hypothesis: Training Deep GNN with One Node One Receptive field12
The Snowflake Hypothesis is introduced -- a novel paradigm underpinning the concept of ``one node, one receptive field'' that can serve as a universal operator for a range of tasks, and it displays tremendous potential on deep GNNs.
- 10
- Numerical studies of CO separation and enrichment from blast furnace gas by using a CuCl/Y fixed bed10
- 9
- 9
- Human Correspondence Consensus for 3D Object Semantic Understanding9
This paper argues that by providing human labeled correspondences between different objects from the same category instead of explicit semantic labels, one can recover rich semantic information of an object and shows that CorresPondenceNet could not only boost fine-grained understanding of heterogeneous objects but also cross-object registration and partial object matching.
- 8
- 7
- DiffAug: Enhance Unsupervised Contrastive Learning with Domain-Knowledge-Free Diffusion-based Data Augmentation6
DiffAug is introduced-a novel and efficient Diffusion-based data Augmentation technique that aims to ensure that the augmented and original data share a smoothed latent space, which is achieved through diffusion steps.
- 6
- 6
- A Study on Transformer Configuration and Training Objective5
Bamboo, an idea of using deeper and narrower transformer configurations, for masked autoencoder training, is proposed, effective in alleviating the over-smoothing issue in deep transformer training.
- Understanding Pixel-level 2D Image Semantics with 3D Keypoint Knowledge Engine4
This paper proposes a new method on predicting image corresponding semantics in 3D domain and then projecting them back onto 2D images to achieve pixel-level understanding and shows that this method gives comparative and even superior results on standard semantic benchmarks.
- 4
- UKPGAN: Unsupervised KeyPoint GANeration.4
This work proposes UKPGAN, an unsupervised 3D keypoint detector where keypoints are detected so that they could reconstruct the original object shape and preserve the semantic information of objects and align well with human annotated part and keypoint labels.
- 4
- 3
- 2
- Perceptual face inpainting with multicolumn gated convolutional network2
A diversified perceptual Markov random field to search correct feature patches in global rather than local images to improve the global semantic coherence and achieve more effective performance in irregular mask.
- Fine-grained Object Semantic Understanding from Correspondences.2
This paper proposes a method that outputs dense semantic embeddings based on a novel geodesic consistency loss and shows that it could boost the fine-grained understanding of heterogeneous objects and the inference of dense semantic information is possible.
- Boosting LLM via Learning from Data Iteratively and Selectively1
This work proposes to perform instruction tuning by iterative data selection by iteratively updating the complexity score for the top-ranked samples and greedily selecting the ones with the highest complexity-diversity score.
- Phantom Experimental Study on Patient Doses Obtained from 320-Multidetector-Row Computed Tomography in Whole-Brain Perfusion Scan1
This study aimed to precisely evaluate organ dose and effective dose (E) obtained from a 320-multidetector-row computed tomography (CT) scanner in brain perfusion scans and to estimate the conversion factor (k) between E and dose length product (DLP).
- 1
- –
- –
- Lightweight DETR for Small‐Target Defect Detection in Stators and Rotors of Pumped Storage Generators–
EdgeFault‐detection transformer (DETR), a lightweight transformer model that integrates three innovations: (1) dynamic geometric‐photometric augmentation for robustness, (2) a FasterNet backbone with Partial Convolution (PConv) that reduces the number of parameters by 40% (from 20 × 106 to 12 × 106), and (3) cross‐scale small‐object head enhancing defect localisation.
- –
- MERIT: Maximum-normalized Element-wise Ratio for Language Model Large-batch Training–
A novel optimizer, MERIT, which leverages the max-norm to calculate the trust ratio to constrain the max attention logit more effectively and further construct element-wise trust ratios to provide more robust update scaling by focusing on local weight structures is proposed.
- –
- –
- Process and Strength Simulation Test Method in Crankshaft Design–
The paper analyzes the internal combustion engine crankshaft fatigue failure forms and main reason and makes strength vibration analysis through the computer simulation.
- –
- –
- –
- –
Publication data from OpenAlex, with missing venues and authors filled in from Crossref; citation counts are the higher of OpenAlex and Semantic Scholar, last synced 2026-10-12. One-sentence summaries under some papers are written by Semantic Scholar’s model. Citation counts may be lower than on Google Scholar, which indexes more sources.
Report an error
Wrong papers, two people merged into one, or a profile that should not be here? Tell us and we will fix or hide it.