Jianzhao Li

Dr. Jianzhao Li

Xidian University, Xi'an, China
orcid

Research Focus & Metrics

Academic Career Length

7Years (2020-Present)

(Calculated once Scilit publication data is synced: current year − year of first publication + 1)

Representative Publications

View All Publications
19 April 2022 in IEEE Transactions on Image Processing
Verified
Journal Article
Closed access
YZYourun Zhang
Yourun Zhang
Maoguo GongMaoguo Gong
Maoguo Gong
Jianzhao LiJianzhao Li
Jianzhao Li
Mingyang ZhangMingyang Zhang
Mingyang Zhang
Fenlong JiangFenlong Jiang
Fenlong Jiang
HZHongyu Zhao
Hongyu Zhao
Extracting 3D information from a single optical image is very attractive. Recently emerging self-supervised methods can learn depth representations without using ground truth depth maps as training data by transforming the depth prediction task into an image synthesis task. However, existing methods rely on a differentiable bilinear sampler for image synthesis, which results in each pixel in a synthetic image being derived from only four pixels in the source image and causes each pixel in the depth map to perceive only a few pixels in the source image. In addition, when calculating the photometric error between a synthetic image and its corresponding target image, existing methods only consider the photometric error within a small neighborhood of each single pixel and therefore ignore correlations between larger areas, which causes the model to tend to fall into the local optima for small patches. In order to extend the perceptual area of the depth map over the source image, we propose a novel multi-scale method that downsamples the predicted depth map and performs image synthesis at different resolutions, which enables each pixel in the depth map to perceive more pixels in the source image and improves the performance of the model. As for the locality of photometric error, we propose a structural similarity (SSIM) pyramid loss to allow the model to sense the difference between images in multiple areas of different sizes. Experimental results show that our method achieves superior performance on both outdoor and indoor benchmarks.

5 January 2023 in IEEE Transactions on Geoscience and Remote Sensing
Verified
Journal Article
Closed access
Jianzhao LiJianzhao Li
Jianzhao Li
Maoguo GongMaoguo Gong
Maoguo Gong
HLHuilin Liu
Huilin Liu
YZYourun Zhang
Yourun Zhang
Mingyang ZhangMingyang Zhang
Mingyang Zhang
Yue WuYue Wu
Yue Wu
Self-supervised learning is an effective way to solve model collapse for few-shot remote sensing scene classification (FSRSSC). However, most self-supervised contrastive learning auxiliary tasks perform poorly on the high inter-class similarity problem in FSRSSC. Furthermore, it is time-consuming and computationally expensive to obtain the best combination among numerous self-supervised auxiliary tasks. In practical applications, we may encounter difficulties in remote sensing data acquisition and labeling, while most FSRSSC studies only focus on the former. To alleviate the above problems, we propose a multiform ensemble self-supervised learning (MES 2 L) framework for FSRSSC in this paper. Based on the transfer learning-based few-shot scheme, we design a novel global-local contrastive learning auxiliary task to solve the low inter-class separability problem. The self-attention mechanism is designed in the local contrast features to investigate the intrinsic associations between different remote sensing scene objectives. We also present a multiform ensemble enhancement (MEE) training method. Ensemble enhancement involves the concatenation of features extracted from different backbones trained by a combination of multiform self-supervised auxiliary tasks. MEE can not only be regarded as a more straightforward alternative to knowledge distillation, but also can achieve an effective compromise between expensive computational cost and classification accuracy. In addition, we provide two scene classification schemes of inductive and transductive settings, corresponding to solving the difficulties of remote sensing data acquisition and labeling. The proposed network achieves state-of-the-art results on three benchmark FSRSSC datasets. The potential of the MES 2 L framework is also demonstrated in combination with classical meta-learning-based and metric learning-based few-shot algorithms.

8 August 2022 in IEEE Transactions on Geoscience and Remote Sensing
Verified
Journal Article
Closed access
Maoguo GongMaoguo Gong
Maoguo Gong
Jianzhao LiJianzhao Li
Jianzhao Li
YZYourun Zhang
Yourun Zhang
Yue WuYue Wu
Yue Wu
Mingyang ZhangMingyang Zhang
Mingyang Zhang
The few-shot scene classification is dedicated to identifying unseen remote sensing classes when only a very small number of labeled samples are available for reference. Most of the existing few-shot scene classification methods are based on meta-learning and use the episodic learning for training, which lacks the consideration for the utilization of data efficiency. In this article, instead of designing sophisticated meta-learning-based algorithms, we are committed to training a feature extractor with good generalization performance and strong feature extraction capability. Specifically, we propose a novel two-path aggregation attention network with quad-patch data augmentation, called data architecture network (DANet), to solve the problem of few-shot scene classification from both data and architecture aspects. In terms of data, we design a new data augmentation strategy named quad-patch augmentation. We use the characteristics of remote sensing images to chunk and reassemble any existing data, thereby generating pseudo-new data to enrich the training set. In terms of architecture, we present a two-path aggregation attention module that makes it easier for the model to focus on the key clues in a targeted manner. The comparative experiments in natural image datasets and remote sensing image datasets demonstrate the effectiveness of our two innovations. In addition, DANet achieves competitive or state-of-the-art (SOTA) results on three benchmark scene classification datasets.

12 November 2021 in IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing
Verified
Journal Article
Open access
Hao LiHao Li
Hao Li
Jianzhao LiJianzhao Li
Jianzhao Li
YZYue Zhao
Yue Zhao
Maoguo GongMaoguo Gong
Maoguo Gong
YZYujing Zhang
Yujing Zhang
Tongfei LiuTongfei Liu
Tongfei Liu
The classification of image time series has potential significance in the field of land-cover analysis with the increasing number of remote sensing images. The key problem of the classification of image time series is how to transfer the already available knowledge on the source domain to the target domain. Nevertheless, most of the existing methods do not consider the impact of different sample costs on the classifier during transferring. In addition, it is very difficult to collect reliable labeled samples with changed or unchanged categories between the source domain and the target domain in the case of a large number of training samples. In order to alleviate the above problems, we propose a cost-sensitive self-paced learning (CSSPL) framework with adaptive regularization for the classification of image time series in this article. Considering that the costs of different samples cannot be completely equal to the classifier, different cost values are assigned to each type of error first, then we minimize the total cost to give the change detection classifier a preference on the unchanged class, aiming to reduce wrong label propagation from the source to the target image. Besides, an adaptive mixture weight regularizer is designed to automatically assign sample weight based on the loss value of the dataset, which enables more reliable sample weights to be selected for training. Experimental results show that the proposed algorithm provides a set of reliable samples for the training of classifier and achieves a promising improvement on classification accuracy.

1 November 2021 in Applied Soft Computing
Verified
Journal Article
Closed access
Jianzhao LiJianzhao Li
Jianzhao Li
Hao LiHao Li
Hao Li
YLYiting Liu
Yiting Liu
MGMaoguo Gong
Maoguo Gong

4 June 2024 in IEEE Transactions on Neural Networks and Learning Systems
Verified
Journal Article
Closed access
YZYourun Zhang
Yourun Zhang
Maoguo GongMaoguo Gong
Maoguo Gong
Jianzhao LiJianzhao Li
Jianzhao Li
KFKyon Fenng
Kyon Fenng
Mingyang ZhangMingyang Zhang
Mingyang Zhang
The few-shot image classification task is to enable a model to identify novel classes by using only a few labeled samples as references. In general, the more knowledge a model has, the more robust it is when facing novel situations. Although directly introducing large amounts of new training data to acquire more knowledge is an attractive solution, it violates the purpose of few-shot learning with respect to reducing dependence on big data. Another viable option is to enable the model to accumulate knowledge more effectively from existing data, i.e., improve the utilization of existing data. In this article, we propose a new data augmentation method called self-mixup (SM) to assemble different augmented instances of the same image, which facilitates the model to more effectively accumulate knowledge from limited training data. In addition to the utilization of data, few-shot learning faces another challenge related to feature extraction. Specifically, existing metric-based few-shot classification methods rely on comparing the extracted features of the novel classes, but the widely adopted downsampling structures in various networks can lead to feature degradation due to the violation of the sampling theorem, and the degraded features are not conducive to robust classification. To alleviate this problem, we propose a calibration-adaptive downsampling (CADS) that calibrates and utilizes the characteristics of different features, which can facilitate robust feature extraction and benefit classification. By improving data utilization and feature extraction, our method shows superior performance on four widely adopted few-shot classification datasets.