Is this you? Claim this profile to correct it, add a bio and choose the work people see first.
Claim this profileWorks54 from public data
- CvT: Introducing Convolutions to Vision Transformers2,569
A new architecture is presented that improves Vision Transformer (ViT) in performance and efficiency by introducing convolutions into ViT to yield the best of both de-signs, and the positional encoding, a crucial component in existing Vision Transformers, can be safely re-moved in this model.
- Dynamic Convolution: Attention Over Convolution Kernels1,508
By simply using dynamic convolution for the state-of-the-art architecture MobileNetV3-Small, the top-1 accuracy of ImageNet classification is boosted by 2.9% with only 4% additional FLOPs and 2.8 AP gain is achieved on COCO keypoint detection.
- Dynamic Head: Unifying Object Detection Heads with Attentions1,044
This paper presents a novel dynamic head framework to unify object detection heads with attentions by coherently combining multiple self-attention mechanisms between feature levels for scale- awareness, among spatial locations for spatial-awareness, and within output channels for task-awareness that significantly improves the representation ability of object detection Heads without any computational overhead.
- RegionCLIP: Region-based Language-Image Pretraining1,000
A new method is proposed that signifi-cantly extends CLIP to learn region-level visual representations, thus enabling fine-grained alignment between image regions and textual concepts, and the learned region representations support zero-shot inference for object detection, showing promising results on both COCO and LVIS datasets.
- Mobile-Former: Bridging MobileNet and Transformer727
The proposed light-weight cross attention to model the bridge enables bidirectional fusion of local and global features and builds an efficient end-to-end detector by replacing backbone, encoder and decoder in DETR with Mobile-Former, which outperforms DETR but saves 52% of computational cost and 36% of parameters.
- Florence-2: Advancing a Unified Representation for a Variety of Vision Tasks657
Florence-2, a novel vision foundation model with a unified, prompt-based representation for various computer vision and vision-language tasks, is introduced and demonstrated to be a strong vision foundation model contender with un-precedented zero-shot and fine-tuning capabilities.
- Rewrite the Stars621
This study attempts to reveal the star operation's ability of mapping inputs into high-dimensional, non-linear feature spaces-akin to kernel tricks-without widening the network.
- 475
- Multi-Scale Vision Longformer: A New Vision Transformer for High-Resolution Image Encoding405
A comprehensive empirical study shows that the new ViT significantly outperforms several strong baselines, including the existing ViT models and their ResNet counterparts, and the Pyramid Vision Transformer from a concurrent work, on a range of vision tasks, including image classification, object detection, and segmentation.
- Generalized Decoding for Pixel, Image, and Language404
X-Decoder is the first work that provides a unified way to support all types of image segmentation and a variety of vision-language (VL) tasks with strong transferability to a wide range of downstream tasks in both zero-shot and finetuning settings.
- MAN: Moment Alignment Network for Natural Language Moment Retrieval via Iterative Graph Adjustment351
This paper presents Moment Alignment Network (MAN), a novel framework that unifies the candidate moment encoding and temporal structural reasoning in a single-shot feed-forward network and proposes to explicitly model moment-wise temporal relations as a structured graph and devise an iterative graph adjustment network to jointly learn the best structure in an end-to-end manner.
- 326
- BEVT: BERT Pretraining of Video Transformers279
BEVT is introduced which decouples video representation learning into spatial representation learning and temporal dynamics learning and achieves state-of-the-art performance on three challenging video benchmarks where BEVT achieves very promising results.
- Temporal Context Network for Activity Localization in Videos271
A Temporal Context Network (TCN) for precise temporal localization of human activities which outperforms state-of-the-art methods on the ActivityNet dataset and the THU-MOS14 dataset.
- 256
- Dynamic ReLU226
This paper proposes dynamic ReLU (DY-ReLU), a dynamic rectifier of which parameters are generated by a hyper function over all in-put elements, and encodes the global context into the hyper function, and adapts the piecewise linear activation function accordingly.
- Masked Video Distillation: Rethinking Masked Feature Modeling for Self-supervised Video Representation Learning135
This paper proposes masked video distillation (MVD), a simple yet effective two-stage masked feature modeling framework for video representation learning that demonstrates that video transformers pre-trained with spatial-temporal co-teaching outperform models distilled with a single teacher on a multitude of video datasets.
- MicroNet: Improving Image Recognition with Extremely Low FLOPs113
This paper found that two factors, sparse connectivity and dynamic activation function, are effective to improve the accuracy and proposes micro-factorized convolution, which factorizes a convolution matrix into low rank matrices, to integrate sparse connectivity into convolution.
- Reduce Information Loss in Transformers for Pluralistic Image Inpainting104
A patch-based auto-encoder P-VQVAE, where the encoder converts the masked image into non-overlapped patch tokens and the decoder recovers the masked regions from the inpainted tokens while keeping the unmasked regions unchanged is designed.
- 82
- ChatVideo: A Tracklet-centric Multimodal and Versatile Video Understanding System79
The vision for multimodal and versatile video understanding is presented and a prototype system, built upon a tracklet-centric paradigm, which treats tracklets as the basic video unit and employs various Video Foundation Models to annotate their properties e.g., appearance, motion, etc.
- 79
- 74
- Look Before You Match: Instance Understanding Matters in Video Object Segmentation61
It is argued that instance understanding matters in VOS, and integrating it with memory-based matching can enjoy the synergy, which is intuitively sensible from the definition of VOS task, i.e., identifying and segmenting object instances within the video.
- Residual Mixture of Experts58
Residual Mixture of Experts (RMoE), an efficient training pipeline for MoE vision transformers on downstream tasks, such as segmentation and detection, achieves comparable results with the upper-bound MoE training, while only introducing minor additional training cost than the lower-bound non-MoE training pipelines.
- OmniTracker: Unifying Visual Object Tracking by Tracking-With-Detection54
A novel tracking-with-detection paradigm is proposed, where tracking supplements appearance priors for detection and detection provides tracking with candidate bounding boxes for the association, and a unified tracking model, OmniTracker, is presented to resolve all the tracking tasks with a fully shared network architecture, model weights, and inference pipeline.
- Dynamic Temporal Pyramid Network: A Closer Look at Multi-scale Modeling for Activity Detection44
The proposed Dynamic Temporal Pyramid Network (DTPN), a new activity detection framework with a multi-scale pyramidal architecture featuring three novel designs, achieves state-of-the-art performance on the challenging ActvityNet dataset.
- 35
- Detection Hub: Unifying Object Detection Datasets via Query Adaptation on Language Embedding33
A new design is addressed by a new design that is dataset-aware and category-aligned that mitigates the dataset inconsistency but also provides coherent guidance for the detector to learn across multiple datasets.
- 27
- TAN: Temporal Aggregation Network for Dense Multi-Label Action Recognition27
Experiments show that the TAN model is well suited for dense multi-label action recognition, which is a challenging sub-topic of action recognition that requires predicting multiple action labels in each frame.
- Layer Grafted Pre-training: Bridging Contrastive Learning And Masked Image Modeling For Label-Efficient Representations25
Inspired by experimental observations, it is found that MIM and CL are suitable to lower and higher layers, respectively, and proposed Layer Grafted Pre-training learns good visual representations that demonstrate superior label efficiency in downstream applications, in particular yielding strong few-shot performance besides linear evaluation.
- Learning from Rich Semantics and Coarse Locations for Long-tailed Object Detection23
This work proposes RichSem, a simple but effective method, which is robust to learn rich semantics from coarse locations without the need of accurate bounding boxes and achieves consistent improvements on both overall and rare-category of LVIS under different backbones and detectors.
- DA-NAS: Data Adapted Pruning for Efficient Neural Architecture Search23
DA-NAS is presented that can directly search the architecture for large-scale target tasks while allowing a large candidate set in a more efficient manner, and supports an argument search space to efficiently search the best-performing architecture.
- 19
- 18
- 18
- 15
- 14
- HyCTAS: Multi-objective hybrid convolution-transformer architecture search for real-time image segmentation12
A series of models via the Hybrid Convolutional-Transformer Architecture Search (HyCTAS) method that searches for the best hybrid combination of lightweight convolution layers and memory-efficient self-attention layers between branches from different resolutions and fuses them at high resolution for both efficiency and effectiveness are presented.
- 12
- 12
- LLM2CLIP: Powerful Language Model Unlocks Richer Cross-Modality Representation11
This work introduces an efficient fine-tuning framework that embeds an LLM into a pretrained CLIP while incurring almost the same training cost as regular CLIP fine-tuning.
- OmniTracker: Unifying Object Tracking by Tracking-with-Detection10
This work proposes a novel tracking-with-detection paradigm, where tracking supplements appearance priors for detection and detection provides tracking with candidate bounding boxes for association and achieves on-par or even better results than both task-specific and unified tracking models.
- 8
- Self-Supervised Learning based on Heat Equation7
A new perspective of self-supervised learning based on extending heat equation into high dimensional feature space is presented, and an insightful hypothesis on the invariance within visual representation over different shapes and textures: the linear relationship between horizontal and vertical derivatives is provided.
- 6
- Video Mobile-Former: Video Recognition with Efficient Global Spatial-temporal Modeling6
Video Mobile-Former is the first Transformer-based video model which constrains the computational budget within 1G FLOPs, and significantly improves the video recognition performance of alternative lightweight baselines, and outperforms other efficient CNN-based models at the low FLOP regime from 500M to 6G totalFLOPs on various video recognition tasks.
- 6
- 3
- 3
- 2
- –
- –
Publication data from OpenAlex; citation counts are the higher of OpenAlex and Semantic Scholar, last synced 2026-10-16. 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.