Is this you? Claim this profile to correct it, add a bio and choose the work people see first.
Claim this profileWorks144 from public data
- GhostNet: More Features From Cheap Operations4,809
A novel Ghost module is proposed to generate more feature maps from cheap operations based on a set of intrinsic feature maps to generate many ghost feature maps that could fully reveal information underlying intrinsic features.
- A Survey on Vision Transformer4,574
This paper reviews these vision transformer models by categorizing them in different tasks and analyzing their advantages and disadvantages, and takes a brief look at the self-attention mechanism in computer vision, as it is the base component in transformer.
- Pre-Trained Image Processing Transformer2,345
To maximally excavate the capability of transformer, the IPT model is presented to utilize the well-known ImageNet benchmark for generating a large amount of corrupted image pairs and the contrastive learning is introduced for well adapting to different image processing tasks.
- CMT: Convolutional Neural Networks Meet Vision Transformers992
A new transformer based hybrid network is proposed by taking advantage of transformers to capture long-range dependencies, and of CNNs to extract local information, obtaining much better trade-off for accuracy and efficiency than previous CNN-based and transformer-based models.
- 964
- DAFL: Data-Free Learning of Student Networks453
A novel framework for training efficient deep neural networks by exploiting generative adversarial networks (GANs) is proposed, where the pre-trained teacher networks are regarded as a fixed discriminator and the generator is utilized for derivating training samples which can obtain the maximum response on the discriminator.
- Data-Free Learning of Student Networks453
A novel framework for training efficient deep neural networks by exploiting generative adversarial networks (GANs) is proposed, where the pre-trained teacher networks are regarded as a fixed discriminator and the generator is utilized for derivating training samples which can obtain the maximum response on the discriminator.
- 399
- 331
- Multimodal Token Fusion for Vision Transformers327
The design of TokenFusion allows the transformer to learn correlations among multimodal features, while the single-modal transformer architecture remains largely intact, and surpasses state-of-the-art methods in three typical vision tasks: multimodals image-to-image translation, RGB-depth semantic segmentation, and 3D object detection with point cloud and images.
- Distilling Object Detectors via Decoupled Features278
This paper presents a novel distillation algorithm via decoupled features (DeFeat) for learning a better student detector that is able to surpass the state-of-the-art distillation methods for object detection.
- CARS: Continuous Evolution for Efficient Neural Architecture Search275
This work develops an efficient continuous evolutionary approach for searching neural networks that provides a series of networks with the number of parameters ranging from 3.7M to 5.1M under mobile settings and surpasses those produced by the state-of-the-art methods on the benchmark ImageNet dataset.
- 248
- AdderNet: Do We Really Need Multiplications in Deep Learning?240
This paper develops a special back-propagation approach for AdderNets by investigating the full-precision gradient, and proposes an adaptive learning rate strategy to enhance the training procedure of Ad DerNets according to the magnitude of each neuron's gradient.
- Patch Slimming for Efficient Vision Transformers218
A novel patch slimming approach that discards useless patches in a topdown paradigm is presented that can significantly reduce the computational costs of vision transformers without affecting their performances.
- CNNpack: packing convolutional neural networks in the frequency domain198
This work presents an effective CNN compression approach in the frequency domain, which focuses not only on smaller weights but on all the weights and their underlying connections, and relax the computational burden of convolution operations in CNNs by linearly combining the convolution responses of discrete cosine transform (DCT) bases.
- 183
- An Image Patch is a Wave: Phase-Aware Vision MLP176
Extensive experiments demonstrate that the proposed Wave-MLP is superior to the state-of-the-art MLP architectures on various vision tasks such as image classification, object detection and semantic segmentation.
- Efficient Residual Dense Block Search for Image Super-Resolution152
An efficient residual dense block search algorithm with multiple objectives to hunt for fast, lightweight and accurate networks for image super-resolution models achieves better performance than the state-of-the-art methods with limited number of parameters and FLOPs.
- 140
- Hire-MLP: Vision MLP via Hierarchical Rearrangement128
Hire-MLP is presented, a simple yet competitive vision MLP architecture via Hierarchical rearrangement, which contains two levels of rearrangements to enable information communication between different regions and capture global context by circularly shifting all tokens along spatial directions.
- 114
- Manifold Regularized Dynamic Network Pruning114
A new paradigm that dynamically removes redundant filters by embedding the manifold information of all instances into the space of pruned networks (dubbed as ManiDP) is proposed, which shows better performance in terms of both accuracy and computational cost compared to the state-of-the-art methods.
- Hit-Detector: Hierarchical Trinity Architecture Search for Object Detection110
This work proposes a hierarchical trinity search framework to simultaneously discover efficient architectures for all components of object detector in an end-to-end manner and empirically reveals that different parts of the detector prefer different operators.
- ReNAS: Relativistic Evaluation of Neural Architecture Search107
This paper proposes a relativistic architecture performance predictor in NAS (ReNAS), encoding neural architectures into feature tensors, and further refining the representations with the predictor, to determine which architecture would perform better instead of accurately predict the absolute architecture performance.
- AdderSR: Towards Energy Efficient Image Super-Resolution100
This paper thoroughly analyzes the relationship between an adder operation and the identity mapping and insert shortcuts to enhance the performance of SR models using adder networks and develops a learnable power activation for adjusting the feature distribution and refining details.
- 97
- Learning Frequency-aware Dynamic Network for Efficient Super-Resolution95
This paper explores a novel frequency-aware dynamic network for dividing the input into multiple parts according to its coefficients in the discrete cosine transform (DCT) domain and can be employed for various SISR neural architectures to obtain the better tradeoff between visual quality and computational complexity.
- 93
- Co-Evolutionary Compression for Unpaired Image Translation93
A novel co-evolutionary approach for reducing their memory usage and FLOPs simultaneously and synergistically optimized for investigating the most important convolution filters iteratively is developed.
- 92
- Distilling Portable Generative Adversarial Networks for Image Translation89
Qualitative and quantitative analysis by conducting experiments on benchmark datasets demonstrate that the proposed method can learn portable generative models with strong performance.
- 87
- UFineBench: Towards Text-based Person Retrieval with Ultra-fine Granularity80
CFAM, a more efficient al-gorithm especially designed for text-based person retrieval with ultra fine-grained texts is proposed, which achieves fine granularity mining by adopting a shared cross-modal granularity de-coder and hard negative match mechanism.
- Masked Image Modeling with Local Multi-Scale Reconstruction77
This model designs local multi-scale reconstruction, where the lower and upper layers reconstruct fine-scale and coarse-scale supervision signals respectively, which accelerates the representation learning process by explicitly guiding multiple layers, but also facilitates multi- scale semantical understanding to the input.
- 76
- Adversarial Learning of Portable Student Networks76
This work utilizes the generative adversarial network (GAN) to learn the student network and demonstrates that the proposed method is capable of learning well-performed portable networks, which is superior to the state-of-the-art methods.
- A Semi-Supervised Assessor of Neural Architectures75
This paper employs an auto-encoder to discover meaningful representations of neural architectures and introduces a graph convolutional neural network to predict the performance of architectures based on the learned representations and their relation modeled by the graph.
- 73
- 66
- Searching for Accurate Binary Neural Architectures65
A new frame work for automatically searching for compact but accurate binary neural networks is presented and it is demonstrated that the searched binary networks can achieve the performance of full-precision models with acceptable increments on model sizes and calculations.
- 59
- One-shot Graph Neural Architecture Search with Dynamic Search Space57
This work proposes a novel dynamic one-shot search space for multi-branch neural architectures of GNNs and demonstrates that this method outperforms the current state-of-the-art manually designed architectures and reaches competitive performance to existing GNN NAS approaches with up to 10 times of speedup.
- 56
- DropNAS: Grouped Operation Dropout for Differentiable Architecture Search56
This paper proposes a novel grouped operation dropout algorithm named DropNAS to fix the problems with DARTS and demonstrates that DropNAS solves the above issues and achieves promising performance.
- Beyond filters: Compact feature map for portable deep model56
This paper focuses on the redundancy in the feature maps derived from the large number of filters in a layer of CNNs, and proposes to extract intrinsic representation of the feature Maps and preserve the discriminability of the features.
- On Positive-Unlabeled Classification in GAN53
A new Positive-Unlabeled GAN is derived that can achieve comparable or even better performance than those sophisticated discriminator stabilization methods and theoretically discusses the global optimality the proposed model will achieve and the equivalent optimization goal.
- Distilling Semantic Priors from SAM to Efficient Image Restoration Models52
This work proposes a general framework to distill SAM's semantic knowledge to boost exiting IR models without interfering with their inference process, and designs a semantic-guided relation (SGR) module for SPD, which ensures semantic feature representation space consistency to fully distill the priors.
- Attribute Aware Pooling for Pedestrian Attribute Recognition52
This paper expands the strength of deep convolutional neural networks to the pedestrian attribute recognition problem by devising a novel attribute aware pooling algorithm that appropriately explores and exploits the correlations between attributes for the pedestrians attribute recognition.
- Learning Versatile Filters for Efficient Convolutional Neural Networks51
Experimental results on benchmark datasets and neural networks demonstrate that CNNs constructed with versatile filters are able to achieve comparable accuracy as that of original filters, but require less memory and FLOPs.
- Towards Evolutionary Compression46
This paper presents an evolutionary method to automatically eliminate redundant convolution filters, representing each compressed network as a binary individual of specific fitness, which has the original network structure and can be directly deployed in any off-the-shelf deep learning libraries.
- 45
- 43
- Beyond Dropout: Feature Map Distortion to Regularize Deep Neural Networks42
This work investigates the empirical Rademacher complexity related to intermediate layers of deep neural networks and proposes a feature distortion method for addressing the problem of over-fitting.
- 41
- 40
- 40
- LegoNet: Efficient Convolutional Neural Networks with Lego Filters39
A split-transform-merge strategy for an efficient convolution by exploiting intermediate Lego feature maps is developed and Inspired by network engineering, it is suggested that an ordinary filter in the neural network can be upgraded to a sophisticated module as well.
- 39
- Residual Distillation: Towards Portable Deep Neural Networks without Shortcuts37
This work proposes a novel joint-training framework to train plain CNN by leveraging the gradients of the ResNet counterpart, and demonstrates that the plainCNN network without shortcuts can achieve the same level of accuracy as that of the ResNet baseline while achieving about 1 .
- 33
- 28
- 28
- 28
- 27
- 24
- 24
- 24
- 23
- 21
- 20
- 20
- 20
- 19
- 18
- 17
- 16
- 15
- 14
- 13
- 11
- 11
- 11
- 10
- 10
- Learning Versatile Convolution Filters for Efficient Visual Recognition9
Experimental results on benchmark datasets and neural networks demonstrate that the versatile filters introduced are able to achieve comparable accuracy as that of original filters, but require less memory and computation cost.
- 9
- 8
- 7
- 7
- 7
- 7
- 6
- 6
- 6
- 6
- 5
- 5
- 5
- 4
- 4
- 4
- 4
- 3
- 3
- 3
- 3
- 3
- 3
- 3
- 3
- 2
- 2
- 2
- 2
- 2
- 1
- 1
- 1
- 1
- 1
- 1
- 1
- 1
- 1
- 1
- 1
- 1
- 1
- 1
- 1
- 1
- ScaleNet: Scaling up Pretrained Neural Networks With Incremental Parameters–
ScaleNet facilitates rapid model expansion with negligible increases in parameters, building on existing pretrained models, and shows significant potential for application in downstream vision areas, as evidenced by the validation in object detection task.
- –
- –
- –
- –
- –
- –
- –
- –
- –
- –
- –
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 2026-10-13. 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.