Qingzhong Wang

Weibin Liao, Haoyi Xiong, Qingzhong Wang et al. · harvard

While self-supervised learning (SSL) algorithms have been widely used to pre-train deep models, few efforts [11] have been done to improve representation learning of X-ray image analysis with SSL pre-trained models. In this work, we study a novel self-supervised pre-training pipeline, namely Multi-task Self-super-vised Continual Learning (MUSCLE), for multiple medical imaging tasks, such as classification and segmentation, using X-ray images collected from multiple body parts, including heads, lungs, and bones. Specifically, MUSCLE aggregates X-rays collected from multiple body parts for MoCo-based representation learning, and adopts a well-designed continual learning (CL) procedure to further pre-train the backbone subject various X-ray analysis tasks jointly. Certain strategies for image pre-processing, learning schedules, and regularization have been used to solve data heterogeneity, overfitting, and catastrophic forgetting problems for multi-task/dataset learning in MUSCLE.We evaluate MUSCLE using 9 real-world X-ray datasets with various tasks, including pneumonia classification, skeletal abnormality classification, lung segmentation, and tuberculosis (TB) detection. Comparisons against other pre-trained models [7] confirm the proof-of-concept that self-supervised multi-task/dataset continual pre-training could boost the performance of X-ray image analysis.

Xueying Zhan, Qingzhong Wang, Kuan-hao Huang et al.

While deep learning (DL) is data-hungry and usually relies on extensive labeled data to deliver good performance, Active Learning (AL) reduces labeling costs by selecting a small proportion of samples from unlabeled data for labeling and training. Therefore, Deep Active Learning (DAL) has risen as a feasible solution for maximizing model performance under a limited labeling cost/budget in recent years. Although abundant methods of DAL have been developed and various literature reviews conducted, the performance evaluation of DAL methods under fair comparison settings is not yet available. Our work intends to fill this gap. In this work, We construct a DAL toolkit, DeepAL+, by re-implementing 19 highly-cited DAL methods. We survey and categorize DAL-related works and construct comparative experiments across frequently used datasets and DAL algorithms. Additionally, we explore some factors (e.g., batch size, number of epochs in the training process) that influence the efficacy of DAL, which provides better references for researchers to design their DAL experiments or carry out DAL-related applications.

Song Zhang, Qingzhong Wang, Jiang Bian et al.

While models derived from Vision Transformers (ViTs) have been phonemically surging, pre-trained models cannot seamlessly adapt to arbitrary resolution images without altering the architecture and configuration, such as sampling the positional encoding, limiting their flexibility for various vision tasks. For instance, the Segment Anything Model (SAM) based on ViT-Huge requires all input images to be resized to 1024$\times$1024. To overcome this limitation, we propose the Multi-Head Self-Attention Convolution (MSA-Conv) that incorporates Self-Attention within generalized convolutions, including standard, dilated, and depthwise ones. Enabling transformers to handle images of varying sizes without retraining or rescaling, the use of MSA-Conv further reduces computational costs compared to global attention in ViT, which grows costly as image size increases. Later, we present the Vision Transformer in Convolution (TiC) as a proof of concept for image classification with MSA-Conv, where two capacity enhancing strategies, namely Multi-Directional Cyclic Shifted Mechanism and Inter-Pooling Mechanism, have been proposed, through establishing long-distance connections between tokens and enlarging the effective receptive field. Extensive experiments have been carried out to validate the overall effectiveness of TiC. Additionally, ablation studies confirm the performance improvement made by MSA-Conv and the two capacity enhancing strategies separately. Note that our proposal aims at studying an alternative to the global attention used in ViT, while MSA-Conv meets our goal by making TiC comparable to state-of-the-art on ImageNet-1K. Code will be released at https://github.com/zs670980918/MSA-Conv.

Yongsong Huang, Qingzhong Wang, Shinichiro Omachi

In this paper, we present a medical AttentIon Denoising Super Resolution Generative Adversarial Network (AID-SRGAN) for diographic image super-resolution. First, we present a medical practical degradation model that considers various degradation factors beyond downsampling. To the best of our knowledge, this is the first composite degradation model proposed for radiographic images. Furthermore, we propose AID-SRGAN, which can simultaneously denoise and generate high-resolution (HR) radiographs. In this model, we introduce an attention mechanism into the denoising module to make it more robust to complicated degradation. Finally, the SR module reconstructs the HR radiographs using the "clean" low-resolution (LR) radiographs. In addition, we propose a separate-joint training approach to train the model, and extensive experiments are conducted to show that the proposed method is superior to its counterparts. e.g., our proposed method achieves $31.90$ of PSNR with a scale factor of $4 \times$, which is $7.05 \%$ higher than that obtained by recent work, SPSR [16]. Our dataset and code will be made available at: https://github.com/yongsongH/AIDSRGAN-MICCAI2022.

Fei Wu, Qingzhong Wang, Jian Bian et al.

To understand human behaviors, action recognition based on videos is a common approach. Compared with image-based action recognition, videos provide much more information. Reducing the ambiguity of actions and in the last decade, many works focused on datasets, novel models and learning approaches have improved video action recognition to a higher level. However, there are challenges and unsolved problems, in particular in sports analytics where data collection and labeling are more sophisticated, requiring sport professionals to annotate data. In addition, the actions could be extremely fast and it becomes difficult to recognize them. Moreover, in team sports like football and basketball, one action could involve multiple players, and to correctly recognize them, we need to analyse all players, which is relatively complicated. In this paper, we present a survey on video action recognition for sports analytics. We introduce more than ten types of sports, including team sports, such as football, basketball, volleyball, hockey and individual sports, such as figure skating, gymnastics, table tennis, tennis, diving and badminton. Then we compare numerous existing frameworks for sports analysis to present status quo of video action recognition in both team sports and individual sports. Finally, we discuss the challenges and unsolved problems in this area and to facilitate sports analytics, we develop a toolbox using PaddlePaddle, which supports football, basketball, table tennis and figure skating action recognition.

Jiuniu Wang, Wenjia Xu, Qingzhong Wang et al.

Recent image captioning models are achieving impressive results based on popular metrics, i.e., BLEU, CIDEr, and SPICE. However, focusing on the most popular metrics that only consider the overlap between the generated captions and human annotation could result in using common words and phrases, which lacks distinctiveness, i.e., many similar images have the same caption. In this paper, we aim to improve the distinctiveness of image captions via comparing and reweighting with a set of similar images. First, we propose a distinctiveness metric -- between-set CIDEr (CIDErBtw) to evaluate the distinctiveness of a caption with respect to those of similar images. Our metric reveals that the human annotations of each image in the MSCOCO dataset are not equivalent based on distinctiveness; however, previous works normally treat the human annotations equally during training, which could be a reason for generating less distinctive captions. In contrast, we reweight each ground-truth caption according to its distinctiveness during training. We further integrate a long-tailed weight strategy to highlight the rare words that contain more information, and captions from the similar image set are sampled as negative examples to encourage the generated sentence to be unique. Finally, extensive experiments are conducted, showing that our proposed approach significantly improves both distinctiveness (as measured by CIDErBtw and retrieval metrics) and accuracy (e.g., as measured by CIDEr) for a wide variety of image captioning baselines. These results are further confirmed through a user study.

Jiang Bian, Xuhong Li, Tao Wang et al.

While deep learning has been widely used for video analytics, such as video classification and action detection, dense action detection with fast-moving subjects from sports videos is still challenging. In this work, we release yet another sports video benchmark \TheName{} for \emph{\underline{P}}ing \emph{\underline{P}}ong-\emph{\underline{A}}ction detection, which consists of 2,721 video clips collected from the broadcasting videos of professional table tennis matches in World Table Tennis Championships and Olympiads. We work with a crew of table tennis professionals and referees on a specially designed annotation toolbox to obtain fine-grained action labels (in 14 classes) for every ping-pong action that appeared in the dataset, and formulate two sets of action detection problems -- \emph{action localization} and \emph{action recognition}. We evaluate a number of commonly-seen action recognition (e.g., TSM, TSN, Video SwinTransformer, and Slowfast) and action localization models (e.g., BSN, BSN++, BMN, TCANet), using \TheName{} for both problems, under various settings. These models can only achieve 48\% area under the AR-AN curve for localization and 82\% top-one accuracy for recognition since the ping-pong actions are dense with fast-moving subjects but broadcasting videos are with only 25 FPS. The results confirm that \TheName{} is still a challenging task and can be used as a special benchmark for dense action detection from videos.

Yuxuan Zhang, Qingzhong Wang, Jiang Bian et al.

To address the problem of medical image recognition, computer vision techniques like convolutional neural networks (CNN) are frequently used. Recently, 3D CNN-based models dominate the field of magnetic resonance image (MRI) analytics. Due to the high similarity between MRI data and videos, we conduct extensive empirical studies on video recognition techniques for MRI classification to answer the questions: (1) can we directly use video recognition models for MRI classification, (2) which model is more appropriate for MRI, (3) are the common tricks like data augmentation in video recognition still useful for MRI classification? Our work suggests that advanced video techniques benefit MRI classification. In this paper, four datasets of Alzheimer's and Parkinson's disease recognition are utilized in experiments, together with three alternative video recognition models and data augmentation techniques that are frequently applied to video tasks. In terms of efficiency, the results reveal that the video framework performs better than 3D-CNN models by 5% - 11% with 50% - 66% less trainable parameters. This report pushes forward the potential fusion of 3D medical imaging and video understanding research.

Xueying Zhan, Zeyu Dai, Qingzhong Wang et al.

Pool-based Active Learning (AL) has achieved great success in minimizing labeling cost by sequentially selecting informative unlabeled samples from a large unlabeled data pool and querying their labels from oracle/annotators. However, existing AL sampling strategies might not work well in out-of-distribution (OOD) data scenarios, where the unlabeled data pool contains some data samples that do not belong to the classes of the target task. Achieving good AL performance under OOD data scenarios is a challenging task due to the natural conflict between AL sampling strategies and OOD sample detection. AL selects data that are hard to be classified by the current basic classifier (e.g., samples whose predicted class probabilities have high entropy), while OOD samples tend to have more uniform predicted class probabilities (i.e., high entropy) than in-distribution (ID) data. In this paper, we propose a sampling scheme, Monte-Carlo Pareto Optimization for Active Learning (POAL), which selects optimal subsets of unlabeled samples with fixed batch size from the unlabeled data pool. We cast the AL sampling task as a multi-objective optimization problem, and thus we utilize Pareto optimization based on two conflicting objectives: (1) the normal AL data sampling scheme (e.g., maximum entropy), and (2) the confidence of not being an OOD sample. Experimental results show its effectiveness on both classical Machine Learning (ML) and Deep Learning (DL) tasks.

Weibin Liao, Xuhong Li, Qingzhong Wang et al.

While pre-training on object detection tasks, such as Common Objects in Contexts (COCO) [1], could significantly boost the performance of cell segmentation, it still consumes on massive fine-annotated cell images [2] with bounding boxes, masks, and cell types for every cell in every image, to fine-tune the pre-trained model. To lower the cost of annotation, this work considers the problem of pre-training DNN models for few-shot cell segmentation, where massive unlabeled cell images are available but only a small proportion is annotated. Hereby, we propose Cross-domain Unsupervised Pre-training, namely CUPre, transferring the capability of object detection and instance segmentation for common visual objects (learned from COCO) to the visual domain of cells using unlabeled images. Given a standard COCO pre-trained network with backbone, neck, and head modules, CUPre adopts an alternate multi-task pre-training (AMT2) procedure with two sub-tasks -- in every iteration of pre-training, AMT2 first trains the backbone with cell images from multiple cell datasets via unsupervised momentum contrastive learning (MoCo) [3], and then trains the whole model with vanilla COCO datasets via instance segmentation. After pre-training, CUPre fine-tunes the whole model on the cell segmentation task using a few annotated images. We carry out extensive experiments to evaluate CUPre using LIVECell [2] and BBBC038 [4] datasets in few-shot instance segmentation settings. The experiment shows that CUPre can outperform existing pre-training methods, achieving the highest average precision (AP) for few-shot cell segmentation and detection.

Yuehai Chen, Qingzhong Wang, Jing Yang et al.

Crowd counting models in highly congested areas confront two main challenges: weak localization ability and difficulty in differentiating between foreground and background, leading to inaccurate estimations. The reason is that objects in highly congested areas are normally small and high level features extracted by convolutional neural networks are less discriminative to represent small objects. To address these problems, we propose a learning discriminative features framework for crowd counting, which is composed of a masked feature prediction module (MPM) and a supervised pixel-level contrastive learning module (CLM). The MPM randomly masks feature vectors in the feature map and then reconstructs them, allowing the model to learn about what is present in the masked regions and improving the model's ability to localize objects in high density regions. The CLM pulls targets close to each other and pushes them far away from background in the feature space, enabling the model to discriminate foreground objects from background. Additionally, the proposed modules can be beneficial in various computer vision tasks, such as crowd counting and object detection, where dense scenes or cluttered environments pose challenges to accurate localization. The proposed two modules are plug-and-play, incorporating the proposed modules into existing models can potentially boost their performance in these scenarios.

Song Zhang, Qingzhong Wang, Junyi Liu et al.

In the fast-growing field of Remote Sensing (RS) image analysis, the gap between massive unlabeled datasets and the ability to fully utilize these datasets for advanced RS analytics presents a significant challenge. To fill the gap, our work introduces an innovative auto-labeling framework named ALPS (Automatic Labeling for Pre-training in Segmentation), leveraging the Segment Anything Model (SAM) to predict precise pseudo-labels for RS images without necessitating prior annotations or additional prompts. The proposed pipeline significantly reduces the labor and resource demands traditionally associated with annotating RS datasets. By constructing two comprehensive pseudo-labeled RS datasets via ALPS for pre-training purposes, our approach enhances the performance of downstream tasks across various benchmarks, including iSAID and ISPRS Potsdam. Experiments demonstrate the effectiveness of our framework, showcasing its ability to generalize well across multiple tasks even under the scarcity of extensively annotated datasets, offering a scalable solution to automatic segmentation and annotation challenges in the field. In addition, the proposed a pipeline is flexible and can be applied to medical image segmentation, remarkably boosting the performance. Note that ALPS utilizes pre-trained SAM to semi-automatically annotate RS images without additional manual annotations. Though every component in the pipeline has bee well explored, integrating clustering algorithms with SAM and novel pseudo-label alignment significantly enhances RS segmentation, as an off-the-shelf tool for pre-training data preparation. Our source code is available at: https://github.com/StriveZs/ALPS.

Qingzhong Wang, Pengfei Zhang, Haoyi Xiong et al.

In this paper, we develop face.evoLVe -- a comprehensive library that collects and implements a wide range of popular deep learning-based methods for face recognition. First of all, face.evoLVe is composed of key components that cover the full process of face analytics, including face alignment, data processing, various backbones, losses, and alternatives with bags of tricks for improving performance. Later, face.evoLVe supports multi-GPU training on top of different deep learning platforms, such as PyTorch and PaddlePaddle, which facilitates researchers to work on both large-scale datasets with millions of images and low-shot counterparts with limited well-annotated data. More importantly, along with face.evoLVe, images before & after alignment in the common benchmark datasets are released with source codes and trained models provided. All these efforts lower the technical burdens in reproducing the existing methods for comparison, while users of our library could focus on developing advanced approaches more efficiently. Last but not least, face.evoLVe is well designed and vibrantly evolving, so that new face recognition approaches can be easily plugged into our framework. Note that we have used face.evoLVe to participate in a number of face recognition competitions and secured the first place. The version that supports PyTorch is publicly available at https://github.com/ZhaoJ9014/face.evoLVe.PyTorch and the PaddlePaddle version is available at https://github.com/ZhaoJ9014/face.evoLVe.PyTorch/tree/master/paddle. Face.evoLVe has been widely used for face analytics, receiving 2.4K stars and 622 forks.

Yaohou Fan, Qingzhong Wang, Yongsong Huang et al.

Current visual text generation models struggle with the trade-off between text accuracy and overall image coherence. We find that achieving high text accuracy can reduce aesthetic quality and instruction-following capability. Although reinforcement learning approaches can alleviate the problem through aligning with multiple rewards, they are often unstable for text generation, as existing approaches normally optimize multiple rewards in a weighted-sum way. In addition, it is difficult to balance the weight of each reward. Moreover, reinforcement learning requires a set of training instructions. A large number of prompts require more training time and computing resources, while a small set leads to poor performance. Hence, how to select the prompts for efficient training is an unsolved problem. In this study, we propose Pareto-Optimal Curriculum Alignment (POCA), a framework that addresses this issue as a multi-objective problem by: 1) identifying the Pareto-optimal set to avoid simple scalarization and 2) designing an adaptive curriculum alignment strategy to manage a learning sequence of a multi-reward dataset using automatic difficulty assessment, which is crucial for optimal convergence as RL methods explore in a limited data environment. In synergy, POCA finds the Pareto-optimal set in a unified reward space, which eliminates inconsistent signals to find the best trade-off solution from different rewards under an easy-to-hard optimization landscape. The experimental results show that POCA significantly improves all metrics such as CLIP, HPS scores and sentence accuracy.

Jiuniu Wang, Wenjia Xu, Qingzhong Wang et al.

Recent advances in image captioning have focused on enhancing accuracy by substantially increasing the dataset and model size. While conventional captioning models exhibit high performance on established metrics such as BLEU, CIDEr, and SPICE, the capability of captions to distinguish the target image from other similar images is under-explored. To generate distinctive captions, a few pioneers employed contrastive learning or re-weighted the ground-truth captions. However, these approaches often overlook the relationships among objects in a similar image group (e.g., items or properties within the same album or fine-grained events). In this paper, we introduce a novel approach to enhance the distinctiveness of image captions, namely Group-based Differential Distinctive Captioning Method, which visually compares each image with other images in one similar group and highlights the uniqueness of each image. In particular, we introduce a Group-based Differential Memory Attention (GDMA) module, designed to identify and emphasize object features in an image that are uniquely distinguishable within its image group, i.e., those exhibiting low similarity with objects in other images. This mechanism ensures that such unique object features are prioritized during caption generation for the image, thereby enhancing the distinctiveness of the resulting captions. To further refine this process, we select distinctive words from the ground-truth captions to guide both the language decoder and the GDMA module. Additionally, we propose a new evaluation metric, the Distinctive Word Rate (DisWordRate), to quantitatively assess caption distinctiveness. Quantitative results indicate that the proposed method significantly improves the distinctiveness of several baseline models, and achieves state-of-the-art performance on distinctiveness while not excessively sacrificing accuracy...

Yihang Yin, Qingzhong Wang, Siyu Huang et al.

Contrastive learning has been widely applied to graph representation learning, where the view generators play a vital role in generating effective contrastive samples. Most of the existing contrastive learning methods employ pre-defined view generation methods, e.g., node drop or edge perturbation, which usually cannot adapt to input data or preserve the original semantic structures well. To address this issue, we propose a novel framework named Automated Graph Contrastive Learning (AutoGCL) in this paper. Specifically, AutoGCL employs a set of learnable graph view generators orchestrated by an auto augmentation strategy, where every graph view generator learns a probability distribution of graphs conditioned by the input. While the graph view generators in AutoGCL preserve the most representative structures of the original graph in generation of every contrastive sample, the auto augmentation learns policies to introduce adequate augmentation variances in the whole contrastive learning procedure. Furthermore, AutoGCL adopts a joint training strategy to train the learnable view generators, the graph encoder, and the classifier in an end-to-end manner, resulting in topological heterogeneity yet semantic similarity in the generation of contrastive samples. Extensive experiments on semi-supervised learning, unsupervised learning, and transfer learning demonstrate the superiority of our AutoGCL framework over the state-of-the-arts in graph contrastive learning. In addition, the visualization results further confirm that the learnable view generators can deliver more compact and semantically meaningful contrastive samples compared against the existing view generation methods.

Yongsong Huang, Zetao Jiang, Qingzhong Wang et al.

Image super-resolution is important in many fields, such as surveillance and remote sensing. However, infrared (IR) images normally have low resolution since the optical equipment is relatively expensive. Recently, deep learning methods have dominated image super-resolution and achieved remarkable performance on visible images; however, IR images have received less attention. IR images have fewer patterns, and hence, it is difficult for deep neural networks (DNNs) to learn diverse features from IR images. In this paper, we present a framework that employs heterogeneous convolution and adversarial training, namely, heterogeneous kernel-based super-resolution Wasserstein GAN (HetSRWGAN), for IR image super-resolution. The HetSRWGAN algorithm is a lightweight GAN architecture that applies a plug-and-play heterogeneous kernel-based residual block. Moreover, a novel loss function that employs image gradients is adopted, which can be applied to an arbitrary model. The proposed HetSRWGAN achieves consistently better performance in both qualitative and quantitative evaluations. According to the experimental results, the whole training process is more stable.

Jiuniu Wang, Wenjia Xu, Qingzhong Wang et al.

Describing images using natural language is widely known as image captioning, which has made consistent progress due to the development of computer vision and natural language generation techniques. Though conventional captioning models achieve high accuracy based on popular metrics, i.e., BLEU, CIDEr, and SPICE, the ability of captions to distinguish the target image from other similar images is under-explored. To generate distinctive captions, a few pioneers employ contrastive learning or re-weighted the ground-truth captions, which focuses on one single input image. However, the relationships between objects in a similar image group (e.g., items or properties within the same album or fine-grained events) are neglected. In this paper, we improve the distinctiveness of image captions using a Group-based Distinctive Captioning Model (GdisCap), which compares each image with other images in one similar group and highlights the uniqueness of each image. In particular, we propose a group-based memory attention (GMA) module, which stores object features that are unique among the image group (i.e., with low similarity to objects in other images). These unique object features are highlighted when generating captions, resulting in more distinctive captions. Furthermore, the distinctive words in the ground-truth captions are selected to supervise the language decoder and GMA. Finally, we propose a new evaluation metric, distinctive word rate (DisWordRate) to measure the distinctiveness of captions. Quantitative results indicate that the proposed method significantly improves the distinctiveness of several baseline models, and achieves the state-of-the-art performance on both accuracy and distinctiveness. Results of a user study agree with the quantitative evaluation and demonstrate the rationality of the new metric DisWordRate.

Siyu Huang, Haoyi Xiong, Zhi-Qi Cheng et al.

Generation of high-quality person images is challenging, due to the sophisticated entanglements among image factors, e.g., appearance, pose, foreground, background, local details, global structures, etc. In this paper, we present a novel end-to-end framework to generate realistic person images based on given person poses and appearances. The core of our framework is a novel generator called Appearance-aware Pose Stylizer (APS) which generates human images by coupling the target pose with the conditioned person appearance progressively. The framework is highly flexible and controllable by effectively decoupling various complex person image factors in the encoding phase, followed by re-coupling them in the decoding phase. In addition, we present a new normalization method named adaptive patch normalization, which enables region-specific normalization and shows a good performance when adopted in person image generation model. Experiments on two benchmark datasets show that our method is capable of generating visually appealing and realistic-looking results using arbitrary image and pose inputs.

Jiuniu Wang, Wenjia Xu, Qingzhong Wang et al.

A wide range of image captioning models has been developed, achieving significant improvement based on popular metrics, such as BLEU, CIDEr, and SPICE. However, although the generated captions can accurately describe the image, they are generic for similar images and lack distinctiveness, i.e., cannot properly describe the uniqueness of each image. In this paper, we aim to improve the distinctiveness of image captions through training with sets of similar images. First, we propose a distinctiveness metric -- between-set CIDEr (CIDErBtw) to evaluate the distinctiveness of a caption with respect to those of similar images. Our metric shows that the human annotations of each image are not equivalent based on distinctiveness. Thus we propose several new training strategies to encourage the distinctiveness of the generated caption for each image, which are based on using CIDErBtw in a weighted loss function or as a reinforcement learning reward. Finally, extensive experiments are conducted, showing that our proposed approach significantly improves both distinctiveness (as measured by CIDErBtw and retrieval metrics) and accuracy (e.g., as measured by CIDEr) for a wide variety of image captioning baselines. These results are further confirmed through a user study.

Di Hu, Lichao Mou, Qingzhong Wang et al.

Visual crowd counting has been recently studied as a way to enable people counting in crowd scenes from images. Albeit successful, vision-based crowd counting approaches could fail to capture informative features in extreme conditions, e.g., imaging at night and occlusion. In this work, we introduce a novel task of audiovisual crowd counting, in which visual and auditory information are integrated for counting purposes. We collect a large-scale benchmark, named auDiovISual Crowd cOunting (DISCO) dataset, consisting of 1,935 images and the corresponding audio clips, and 170,270 annotated instances. In order to fuse the two modalities, we make use of a linear feature-wise fusion module that carries out an affine transformation on visual and auditory features. Finally, we conduct extensive experiments using the proposed dataset and approach. Experimental results show that introducing auditory information can benefit crowd counting under different illumination, noise, and occlusion conditions. The dataset and code will be released. Code and data have been made available

Siyu Huang, Haoyi Xiong, Tianyang Wang et al.

Arbitrary image style transfer is a challenging task which aims to stylize a content image conditioned on arbitrary style images. In this task the feature-level content-style transformation plays a vital role for proper fusion of features. Existing feature transformation algorithms often suffer from loss of content or style details, non-natural stroke patterns, and unstable training. To mitigate these issues, this paper proposes a new feature-level style transformation technique, named Style Projection, for parameter-free, fast, and effective content-style transformation. This paper further presents a real-time feed-forward model to leverage Style Projection for arbitrary image style transfer, which includes a regularization term for matching the semantics between input contents and stylized outputs. Extensive qualitative analysis, quantitative evaluation, and user study have demonstrated the effectiveness and efficiency of the proposed methods.

Qingzhong Wang, Antoni B. Chan

Although significant progress has been made in the field of automatic image captioning, it is still a challenging task. Previous works normally pay much attention to improving the quality of the generated captions but ignore the diversity of captions. In this paper, we combine determinantal point process (DPP) and reinforcement learning (RL) and propose a novel reinforcing DPP (R-DPP) approach to generate a set of captions with high quality and diversity for an image. We show that R-DPP performs better on accuracy and diversity than using noise as a control signal (GANs, VAEs). Moreover, R-DPP is able to preserve the modes of the learned distribution. Hence, beam search algorithm can be applied to generate a single accurate caption, which performs better than other RL-based models.

Qingzhong Wang, Antoni B. Chan

Recently, the state-of-the-art models for image captioning have overtaken human performance based on the most popular metrics, such as BLEU, METEOR, ROUGE, and CIDEr. Does this mean we have solved the task of image captioning? The above metrics only measure the similarity of the generated caption to the human annotations, which reflects its accuracy. However, an image contains many concepts and multiple levels of detail, and thus there is a variety of captions that express different concepts and details that might be interesting for different humans. Therefore only evaluating accuracy is not sufficient for measuring the performance of captioning models --- the diversity of the generated captions should also be considered. In this paper, we proposed a new metric for measuring the diversity of image captions, which is derived from latent semantic analysis and kernelized to use CIDEr similarity. We conduct extensive experiments to re-evaluate recent captioning models in the context of both diversity and accuracy. We find that there is still a large gap between the model and human performance in terms of both accuracy and diversity and the models that have optimized accuracy (CIDEr) have low diversity. We also show that balancing the cross-entropy loss and CIDEr reward in reinforcement learning during training can effectively control the tradeoff between diversity and accuracy of the generated captions.

Qingzhong Wang, Antoni B. Chan

Attention modules connecting encoder and decoders have been widely applied in the field of object recognition, image captioning, visual question answering and neural machine translation, and significantly improves the performance. In this paper, we propose a bottom-up gated hierarchical attention (GHA) mechanism for image captioning. Our proposed model employs a CNN as the decoder which is able to learn different concepts at different layers, and apparently, different concepts correspond to different areas of an image. Therefore, we develop the GHA in which low-level concepts are merged into high-level concepts and simultaneously low-level attended features pass to the top to make predictions. Our GHA significantly improves the performance of the model that only applies one level attention, for example, the CIDEr score increases from 0.923 to 0.999, which is comparable to the state-of-the-art models that employ attributes boosting and reinforcement learning (RL). We also conduct extensive experiments to analyze the CNN decoder and our proposed GHA, and we find that deeper decoders cannot obtain better performance, and when the convolutional decoder becomes deeper the model is likely to collapse during training.

Qingzhong Wang, Antoni B. Chan

Image captioning is a challenging task that combines the field of computer vision and natural language processing. A variety of approaches have been proposed to achieve the goal of automatically describing an image, and recurrent neural network (RNN) or long-short term memory (LSTM) based models dominate this field. However, RNNs or LSTMs cannot be calculated in parallel and ignore the underlying hierarchical structure of a sentence. In this paper, we propose a framework that only employs convolutional neural networks (CNNs) to generate captions. Owing to parallel computing, our basic model is around 3 times faster than NIC (an LSTM-based model) during training time, while also providing better results. We conduct extensive experiments on MSCOCO and investigate the influence of the model width and depth. Compared with LSTM-based models that apply similar attention mechanisms, our proposed models achieves comparable scores of BLEU-1,2,3,4 and METEOR, and higher scores of CIDEr. We also test our model on the paragraph annotation dataset, and get higher CIDEr score compared with hierarchical LSTMs