A5036993002- Emotion and Mood Recognition
- Face and Expression Recognition
- Face recognition and analysis
- Advanced Image Processing Techniques
- Speech and Audio Processing
- Image and Signal Denoising Methods
- Image Processing Techniques and Applications
- Image Enhancement Techniques
- Advanced Vision and Imaging
- Anomaly Detection Techniques and Applications
- Advanced Neural Network Applications
- Advanced Image Fusion Techniques
- Adversarial Robustness in Machine Learning
- Generative Adversarial Networks and Image Synthesis
- Advanced Image and Video Retrieval Techniques
- Video Surveillance and Tracking Methods
- Image and Object Detection Techniques
- Advanced Fiber Laser Technologies
- Target Tracking and Data Fusion in Sensor Networks
- EEG and Brain-Computer Interfaces
- Down syndrome and intellectual disability research
- Laser-Matter Interactions and Applications
- Silicon and Solar Cell Technologies
- Infrared Target Detection Methodologies
- Solid State Laser Technologies
Hebei University of Technology
2022
Nanchang University
2021
China Mobile (China)
2021
Peking University
2020
University of South Carolina
2014-2019
Huazhong University of Science and Technology
2019
Zhejiang University
2012
A training process for facial expression recognition is usually performed sequentially in three individual stages: feature learning, selection, and classifier construction. Extensive empirical studies are needed to search an optimal combination of representation, set, achieve good performance. This paper presents a novel Boosted Deep Belief Network (BDBN) performing the stages iteratively unified loopy framework. Through proposed BDBN framework, set features, which effective characterize...
Facial expression recognition suffers under realworldconditions, especially on unseen subjects due to highinter-subject variations. To alleviate variations introduced bypersonal attributes and achieve better facial recognitionperformance, a novel identity-aware convolutional neuralnetwork (IACNN) is proposed. In particular, CNN with newarchitecture employed as individual streams of bi-streamidentity-aware network. An expression-sensitive contrastive lossis developed measure the similarity...
Over the past few years, Convolutional Neural Networks (CNNs) have shown promise on facial expression recognition. However, performance degrades dramatically under real-world settings due to variations introduced by subtle appearance changes, head pose variations, illumination and occlusions. In this paper, a novel island loss is proposed enhance discriminative power of deeply learned features. Specifically, designed reduce intra-class while enlarging inter-class differences simultaneously....
This paper reviews the NTIRE 2020 challenge on real world super-resolution. It focuses participating methods and final results. The addresses setting, where paired true high low-resolution images are unavailable. For training, only one set of source input is therefore provided along with a unpaired high-quality target images. In Track 1: Image Processing artifacts, aim to super-resolve synthetically generated image processing artifacts. allows for quantitative benchmarking approaches w.r.t....
Video super-resolution has recently become one of the most important mobile-related problems due to rise video communication and streaming services. While many solutions have been proposed for this task, majority them are too computationally expensive run on portable devices with limited hardware resources. To address problem, we introduce first Mobile AI challenge, where target is develop an end-to-end deep learning-based that can achieve a real-time performance mobile GPUs. The...
Recognizing facial action units (AUs) during spontaneous displays is a challenging problem. Most recently, Convolutional Neural Networks (CNNs) have shown promise for AU recognition, where predefined and fixed convolution filter sizes are employed. In order to achieve the best performance, optimal size often empirically found by conducting extensive experimental validation. Such training process suffers from expensive cost, especially as network becomes deeper. This paper proposes novel...
A novel Identity-Free conditional Generative Adversarial Network (IF-GAN) was proposed for Facial Expression Recognition (FER) to explicitly reduce high inter-subject variations caused by identity-related facial attributes, e.g., age, race, and gender. As part of an end-to-end system, a cGAN designed transform given input expression "average" identity face with the same as input. Then, identity-free FER is possible since generated images have synthetic differ only in their displayed...
Detecting the salient objects in a remote sensing image has wide applications. Many existing deep learning methods have been proposed for Salient Object Detection (SOD) images with remarkable results. However, recent adversarial attack examples, generated by changing few pixel values on original image, could result collapse well-trained model. Different adding perturbation to images, we propose jointly tune exposure and additive constrain close cloudy as Adversarial Cloud. Cloud is natural...
In the past decade, Convolutional Neural Networks (CNNs) have been demonstrated successful for object detections. However, size of network input is limited by amount memory available on GPUs. Moreover, performance degrades when detecting small objects. To alleviate usage and improve traffic signs, we proposed an approach signs from large images under real world conditions. particular, are broken into patches as to a Small-Object-Sensitive-CNN (SOS-CNN) modified Single Shot Multibox Detector...
Recognizing facial action units (AUs) from spontaneous expressions is still a challenging problem. Most recently, CNNs have shown promise on AU recognition. However, the learned are often overfitted and do not generalize well to unseen subjects due limited AU-coded training images. We proposed novel Incremental Boosting CNN (IB-CNN) integrate boosting into via an incremental layer that selects discriminative neurons lower incrementally updated successive mini-batches. In addition, loss...
The phenomenon of reflection is quite common in digital images, posing significant challenges for various applications such as computer vision, photography, and image processing. Traditional methods removal often struggle to achieve clean results while maintaining high fidelity robustness, particularly real-world scenarios. Over the past few decades, numerous deep learning-based approaches have emerged, yielding impressive results. In this survey, we conduct a comprehensive review current...
A novel Identity-Free conditional Generative Adversarial Network (IF-GAN) was proposed for Facial Expression Recognition (FER) to explicitly reduce high inter-subject variations caused by identity-related facial attributes, e.g., age, race, and gender. As part of an end-to-end system, a cGAN designed transform given input expression image "average" identity face with the same as input. Then, identity-free FER is possible since generated images have synthetic differ only in their displayed...
In this article, we propose a simple yet effective approach, called point adversarial self mining (PASM), to improve the recognition accuracy in facial expression (FER). Unlike previous works focusing on designing specific architectures or loss functions solve problem, PASM boosts network capability by simulating human learning processes: providing updated materials and guidance from more capable teachers. Specifically, generate new materials, leverages attack method trained teacher locate...
Very recent work has demonstrated tremendous improvements in facial expression recognition (FER) on laboratory-controlled datasets. However, recognizing expressions under in-the-wild conditions still remains challenging, especially unseen subjects due to high inter-subject variations. In this paper, we propose a novel Probabilistic Attribute Tree Convolutional Neural Network (PAT-CNN) explicitly deal with large intra-class variations caused by identity-related attributes, e.g., age, race,...
Restoring and inpainting the visual memories that are present, but often impaired, in old photos remains an intriguing unsolved research topic. Decades-old suffer from severe commingled degradation such as cracks, defocus, color-fading, which difficult to treat individually harder repair when they interact. Deep learning presents a plausible avenue, lack of large-scale datasets makes addressing this restoration task very challenging. Here we present novel reference-based end-to-end framework...
We develop a person identification approach - Clothing Change Aware Network (CCAN) for the task of clothing assisted identification. CCAN concerns approaches that go beyond face recognition and particularly tackles role to Person is rather challenging when appears changed under complex background information. With pair two images as input, simultaneously performs verification detect change in an predict identity. When from input detected be different, automatically understates information...
Emotion recognition plays an important role in human-computer interaction (HCI) and has been extensively studied for decades. Although tremendous improvements have achieved posed expressions, recognizing human emotions "close-to-real-world" environments remains a challenge. In this paper, we proposed two strategies to fuse information extracted from different modalities, i.e., audio visual. Specifically, utilized LBP-TOP, ensemble of CNNs, bi-directional LSTM (BLSTM) extract features the...
Image super-resolution is one of the important computer vision techniques aiming to reconstruct high-resolution images from corresponding low-resolution ones. Most recently, deep learning based approaches have been demonstrated for image super-resolution. However, as networks go deeper, they become more difficult train and restore finer texture details, especially under real-world settings. In this paper, we propose a Residual Channel Attention-Generative Adversarial Network (RCA-GAN) solve...
Group-level Emotion Recognition (GER) in the wild is a challenging task gaining lots of attention. Most recent works utilized two channels information, channel involving only faces and containing whole image, to solve this problem. However, modeling relationship between scene global image remains challenging. In paper, we proposed novel face-location aware network, capturing face location information form an attention heatmap better model such relationships. We also multi-scale network infer...
In this work, we proposed adaptive pooling maps (APMs) for CNNs to aid facial expression recognition. Inspired by superpixels, which represent the image content more naturally, consisting of irregular regions are learned from training images as part a CNN model. The APMs preserve local structural information and thus capable capturing subtle appearance geometrical changes caused expression. Furthermore, developed an efficient algorithm learn efficiently. Experiments on three benchmark...
In this paper, we proposed a novel Probabilistic Attribute Tree-CNN (PAT-CNN) to explicitly deal with the large intra-class variations caused by identity-related attributes, e.g., age, race, and gender. Specifically, PAT module an associated loss was learn features in hierarchical tree structure organized according where final are less affected attributes. Then, expression-related extracted from leaf nodes. Samples probabilistically assigned nodes at different levels such that can be learned...