QGQingji Guan

Qingji Guan

*This is an algorithmically generated author record based on publicly available information.

Current Organizations / Institutes

The organizations/institutes from the scholar's latest article

Previous Organizations / Institutes

The organizations / institutes from the scholar's other articles in the past ten years


Qingji Guan has a total of 23 co-authors. Here are the top co-authors by number of shared publications:


Latest Publications

  • ReFrame: A Resource-Friendly Cloud-Assisted On-Device Deep Learning Framework for Vision Services

    by Jianhang Xie, Chuntao Ding, Qingji Guan, Ao Zhou, Yidong Li

    Cloud-assisted Internet of Things (IoT) device deployment of deep neural networks (DNNs) promotes On-device deep learning to provide users with ubiquitous high-quality services by solving the contradiction between insufficient IoT device resources and intensive demand for high-performance DNN resources. However, most existing methods optimize DNNs by considering one or two terms of transmission, computation, and storage resources, but do not consider all three terms at the same time in cloud-assisted IoT device deployment and updating DNNs. To this end, we propose a non-learnable module-based ResNet and a cloud-assisted on-device deep learning framework, ReFrame, based on the consideration of three indicators: model transmission parameters, computation resources, and storage resources. In the proposed method, we first specify that some parameters in DNNs are non-learnable and randomly initialized, so that, these parameters can be saved and reproduced with a few random seeds. By doing so, the cloud only transmits random seeds and learnable parameters to reduce the number of parameter transmissions. Secondly, we reduce the computation resource consumption of the model by introducing computation-friendly operators, such as pooling, to replace vanilla convolutions. Finally, since random seeds are used to save non-learnable model parameters, on IoT devices we only need to store random seeds and learnable parameters to reproduce the well-trained model. Compared with saving the complete model, our method greatly reduces IoT device storage resource consumption. Experimental results on image classification, object detection, and semantic segmentation tasks demonstrate the effectiveness of the proposed method. Specifically, on the CIFAR-10, our proposed method reduces approximately 89% of FLOPs and 90% of transmitted data in the prototype system compared to ResNet-18.


  • Improving Image Inpainting via Adversarial Collaborative Training

    by Li Huang, Yaping Huang, Qingji Guan

    Image inpainting aims to restore visually realistic contents from a corrupted image, while inpainting forensic methods focus on locating the inpainted regions to fight against inpainting manipulations. Motivated by these two mutually interdependent tasks, in this paper, we propose a novel image inpainting network called Adversarial Collaborative Network (AdvColabNet), which leverages the contradictory and collaborative information from the two tasks of image inpainting and inpainting forensics to enhance the progress of the inpainting model through adversarial collaborative training. Specifically, the proposed AdvColabNet is a coarse-to-fine two-stage framework. In the coarse training stage, a simple generative adversarial model-based U-Net-style network generates initial coarse inpainting results. In the fine stage, the authenticity of inpainting results is assessed using the estimated forensic mask. A forensics-driven adaptive weighting refinement strategy is developed to emphasize learning from pixels with higher probabilities of being inpainted, which helps the network to focus on the challenging regions, resulting in more plausible inpainting results. Comprehensive evaluations on the CelebA-HQ and Places2 datasets demonstrate that our method achieves state-of-the-art robustness performance in terms of PSNR, SSIM, MAE, FID, and LPIPS metrics. We also show that our method effectively deceives the proposed inpainting forensic method compared to state-of-the-art inpainting methods, further demonstrating the superiority of the proposed method.


  • Open Access

    An Improved Heteroscedastic Modeling Method for Chest X-ray Image Classification with Noisy Labels

    by Qingji Guan, Qinrun Chen, Yaping Huang

    Chest X-ray image classification suffers from the high inter-similarity in appearance that is vulnerable to noisy labels. The data-dependent and heteroscedastic characteristic label noise make chest X-ray image classification more challenging. To address this problem, in this paper, we first revisit the heteroscedastic modeling (HM) for image classification with noise labels. Rather than modeling all images in one fell swoop as in HM, we instead propose a novel framework that considers the noisy and clean samples separately for chest X-ray image classification. The proposed framework consists of a Gaussian Mixture Model-based noise detector and a Heteroscedastic Modeling-based noise-aware classification network, named GMM-HM. The noise detector is constructed to judge whether one sample is clean or noisy. The noise-aware classification network models the noisy and clean samples with heteroscedastic and homoscedastic hypotheses, respectively. Through building the correlations between the corrupted noisy samples, the GMM-HM is much more robust than HM, which uses only the homoscedastic hypothesis. Compared with HM, we show consistent improvements on the ChestX-ray2017 dataset with different levels of symmetric and asymmetric noise. Furthermore, we also conduct experiments on a real asymmetric noisy dataset, ChestX-ray14. The experimental results on ChestX-ray14 show the superiority of the proposed method.


  • Open Access

    The Treasure Beneath Multiple Annotations: An Uncertainty-aware Edge Detector

    by Caixia Zhou, Yaping Huang, Mengyang Pu, Qingji Guan, Li Huang, Haibin Ling

    Deep learning-based edge detectors heavily rely on pixel-wise labels which are often provided by multiple annotators. Existing methods fuse multiple annotations using a simple voting process, ignoring the inherent ambiguity of edges and labeling bias of annotators. In this paper, we propose a novel uncertainty-aware edge detector (UAED), which employs uncertainty to investigate the subjectivity and ambiguity of diverse annotations. Specifically, we first convert the deterministic label space into a learnable Gaussian distribution, whose variance measures the degree of ambiguity among different annotations. Then we regard the learned variance as the estimated uncertainty of the predicted edge maps, and pixels with higher uncertainty are likely to be hard samples for edge detection. Therefore we design an adaptive weighting loss to emphasize the learning from those pixels with high uncertainty, which helps the network to gradually concentrate on the important pixels. UAED can be combined with various encoder-decoder backbones, and the extensive experiments demonstrate that UAED achieves superior performance consistently across multiple edge detection benchmarks. The source code is available at \url{https://github.com/ZhouCX117/UAED}

  • Thorax disease classification with attention guided convolutional neural network

    by Qingji Guan, Yaping Huang, Zhun Zhong, Zhedong Zheng, Liang Zheng, Yi Yang


  • Making better use of edges for sketch generation

    by Xingyuan Zhang, Yaping Huang, Qi Zou, Qingji Guan, Junbo Liu

    Sketching has become fashionable with the increasing availability of touch-screens on portable devices. It is typically used for rendering the visual world, automatic sketch style recognition and abstraction, sketch-based image retrieval (SBIR), and sketch-based perceptual grouping. How to automatically generate a sketch from a real image remains an open question. We propose a convolutional neural network-based model, named SG-Net, to generate sketches from natural images. SG-Net is trained to learn the relationship between images and sketches and thus makes full use of edge information to generate a rough sketch. Then, mathematical morphology is further utilized as a postprocess to eliminate the redundant artifacts in the generated sketches. In addition, in order to increase the diversity of generated sketches, we introduce thin plate splines to generate more sketches with different styles. We evaluate the proposed method of sketch generation both quantitatively and qualitatively on the challenging dataset. Our approach achieves superior performance to the established methods. Moreover, we conduct extensive experiments on the SBIR task. The experimental results on the Flickr15k dataset demonstrate that our proposed method leverages the retrieval performance compared with the state-of-the-art methods.