Sun Li

Siyuan Li, Li Sun, Qingli Li

Pre-trained vision-language models like CLIP have recently shown superior performances on various downstream tasks, including image classification and segmentation. However, in fine-grained image re-identification (ReID), the labels are indexes, lacking concrete text descriptions. Therefore, it remains to be determined how such models could be applied to these tasks. This paper first finds out that simply fine-tuning the visual model initialized by the image encoder in CLIP, has already obtained competitive performances in various ReID tasks. Then we propose a two-stage strategy to facilitate a better visual representation. The key idea is to fully exploit the cross-modal description ability in CLIP through a set of learnable text tokens for each ID and give them to the text encoder to form ambiguous descriptions. In the first training stage, image and text encoders from CLIP keep fixed, and only the text tokens are optimized from scratch by the contrastive loss computed within a batch. In the second stage, the ID-specific text tokens and their encoder become static, providing constraints for fine-tuning the image encoder. With the help of the designed loss in the downstream task, the image encoder is able to represent data as vectors in the feature embedding accurately. The effectiveness of the proposed strategy is validated on several datasets for the person or vehicle ReID tasks. Code is available at https://github.com/Syliz517/CLIP-ReID.

Yuzhen Qin, Li Sun, Hui Chen et al.

The widespread emergence of smart devices for ECG has sparked demand for intelligent single-lead ECG-based diagnostic systems. However, it is challenging to develop a single-lead-based ECG interpretation model for multiple diseases diagnosis due to the lack of some key disease information. In this work, we propose inter-lead Multi-View Knowledge Transferring of ECG (MVKT-ECG) to boost single-lead ECG's ability for multi-label disease diagnosis. This training strategy can transfer superior disease knowledge from multiple different views of ECG (e.g. 12-lead ECG) to single-lead-based ECG interpretation model to mine details in single-lead ECG signals that are easily overlooked by neural networks. MVKT-ECG allows this lead variety as a supervision signal within a teacher-student paradigm, where the teacher observes multi-lead ECG educates a student who observes only single-lead ECG. Since the mutual disease information between the single-lead ECG and muli-lead ECG plays a key role in knowledge transferring, we present a new disease-aware Contrastive Lead-information Transferring(CLT) to improve the mutual disease information between the single-lead ECG and muli-lead ECG. Moreover, We modify traditional Knowledge Distillation to multi-label disease Knowledge Distillation (MKD) to make it applicable for multi-label disease diagnosis. The comprehensive experiments verify that MVKT-ECG has an excellent performance in improving the diagnostic effect of single-lead ECG.

Ke Yu, Li Sun, Junxiang Chen et al.

Large-scale volumetric medical images with annotation are rare, costly, and time prohibitive to acquire. Self-supervised learning (SSL) offers a promising pre-training and feature extraction solution for many downstream tasks, as it only uses unlabeled data. Recently, SSL methods based on instance discrimination have gained popularity in the medical imaging domain. However, SSL pre-trained encoders may use many clues in the image to discriminate an instance that are not necessarily disease-related. Moreover, pathological patterns are often subtle and heterogeneous, requiring the ability of the desired method to represent anatomy-specific features that are sensitive to abnormal changes in different body parts. In this work, we present a novel SSL framework, named DrasCLR, for 3D medical imaging to overcome these challenges. We propose two domain-specific contrastive learning strategies: one aims to capture subtle disease patterns inside a local anatomical region, and the other aims to represent severe disease patterns that span larger regions. We formulate the encoder using conditional hyper-parameterized network, in which the parameters are dependant on the anatomical location, to extract anatomically sensitive features. Extensive experiments on large-scale computer tomography (CT) datasets of lung images show that our method improves the performance of many downstream prediction and segmentation tasks. The patient-level representation improves the performance of the patient survival prediction task. We show how our method can detect emphysema subtypes via dense prediction. We demonstrate that fine-tuning the pre-trained model can significantly reduce annotation efforts without sacrificing emphysema detection accuracy. Our ablation study highlights the importance of incorporating anatomical context into the SSL framework.

Li Sun, Ke Yu, Kayhan Batmanghelich

Although self-supervised learning enables us to bootstrap the training by exploiting unlabeled data, the generic self-supervised methods for natural images do not sufficiently incorporate the context. For medical images, a desirable method should be sensitive enough to detect deviation from normal-appearing tissue of each anatomical region; here, anatomy is the context. We introduce a novel approach with two levels of self-supervised representation learning objectives: one on the regional anatomical level and another on the patient-level. We use graph neural networks to incorporate the relationship between different anatomical regions. The structure of the graph is informed by anatomical correspondences between each patient and an anatomical atlas. In addition, the graph representation has the advantage of handling any arbitrarily sized image in full resolution. Experiments on large-scale Computer Tomography (CT) datasets of lung images show that our approach compares favorably to baseline methods that do not account for the context. We use the learned embedding for staging lung tissue abnormalities related to COVID-19.

Yang Li, Guijin Wang, Zhourui Xia et al.

Auxiliary diagnosis of cardiac electrophysiological status can be obtained through the analysis of 12-lead electrocardiograms (ECGs). This work proposes a dual-scale lead-separated transformer with lead-orthogonal attention and meta-information (DLTM-ECG) as a novel approach to address this challenge. ECG segments of each lead are interpreted as independent patches, and together with the reduced dimension signal, they form a dual-scale representation. As a method to reduce interference from segments with low correlation, two group attention mechanisms perform both lead-internal and cross-lead attention. Our method allows for the addition of previously discarded meta-information, further improving the utilization of clinical information. Experimental results show that our DLTM-ECG yields significantly better classification scores than other transformer-based models,matching or performing better than state-of-the-art (SOTA) deep learning methods on two benchmark datasets. Our work has the potential for similar multichannel bioelectrical signal processing and physiological multimodal tasks.

Zhi Yan, Li Sun, Tomas Krajnik et al.

In the future, service robots are expected to be able to operate autonomously for long periods of time without human intervention. Many work striving for this goal have been emerging with the development of robotics, both hardware and software. Today we believe that an important underpinning of long-term robot autonomy is the ability of robots to learn on site and on-the-fly, especially when they are deployed in changing environments or need to traverse different environments. In this paper, we examine the problem of long-term autonomy from the perspective of robot learning, especially in an online way, and discuss in tandem its premise "data" and the subsequent "deployment".

Junyi Fan, Li Sun, Negin Ashrafi et al.

Nursing documentation in intensive care units (ICUs) provides essential clinical intelligence but often suffers from inconsistent terminology, informal styles, and lack of standardization, challenges that are particularly critical in heart failure care. This study applies Direct Preference Optimization (DPO) to adapt Mistral-7B, a locally deployable language model, using 8,838 heart failure nursing notes from the MIMIC-III database and 21,210 preference pairs derived from expert-verified GPT outputs, model generations, and original notes. Evaluation across BLEU, ROUGE, BERTScore, Perplexity, and expert qualitative assessments demonstrates that DPO markedly enhances documentation quality. Specifically, BLEU increased by 84% (0.173 to 0.318), BERTScore improved by 7.6% (0.828 to 0.891), and expert ratings rose across accuracy (+14.4 points), completeness (+14.5 points), logical consistency (+14.1 points), readability (+11.1 points), and structural clarity (+6.0 points). These results indicate that DPO can align lightweight clinical language models with expert standards, supporting privacy-preserving, AI-assisted documentation within electronic health record systems to reduce administrative burden and improve ICU patient safety.

Irum Mehboob, Li Sun, Alireza Astegarpanah et al.

This paper shows how an uncertainty-aware, deep neural network can be trained to detect, recognise and localise objects in 2D RGB images, in applications lacking annotated train-ng datasets. We propose a self-supervising teacher-student pipeline, in which a relatively simple teacher classifier, trained with only a few labelled 2D thumbnails, automatically processes a larger body of unlabelled RGB-D data to teach a student network based on a modified YOLOv3 architecture. Firstly, 3D object detection with back projection is used to automatically extract and teach 2D detection and localisation information to the student network. Secondly, a weakly supervised 2D thumbnail classifier, with minimal training on a small number of hand-labelled images, is used to teach object category recognition. Thirdly, we use a Gaussian Process GP to encode and teach a robust uncertainty estimation functionality, so that the student can output confidence scores with each categorization. The resulting student significantly outperforms the same YOLO architecture trained directly on the same amount of labelled data. Our GP-based approach yields robust and meaningful uncertainty estimations for complex industrial object classifications. The end-to-end network is also capable of real-time processing, needed for robotics applications. Our method can be applied to many important industrial tasks, where labelled datasets are typically unavailable. In this paper, we demonstrate an example of detection, localisation, and object category recognition of nuclear mixed-waste materials in highly cluttered and unstructured scenes. This is critical for robotic sorting and handling of legacy nuclear waste, which poses complex environmental remediation challenges in many nuclearised nations.

Mingyu Yin, Li Sun, Qingli Li

Novel view synthesis often needs the paired data from both the source and target views. This paper proposes a view translation model under cVAE-GAN framework without requiring the paired data. We design a conditional deformable module (CDM) which uses the view condition vectors as the filters to convolve the feature maps of the main branch in VAE. It generates several pairs of displacement maps to deform the features, like the 2D optical flows. The results are fed into the deformed feature based normalization module (DFNM), which scales and offsets the main branch feature, given its deformed one as the input from the side branch. Taking the advantage of the CDM and DFNM, the encoder outputs a view-irrelevant posterior, while the decoder takes the code drawn from it to synthesize the reconstructed and the viewtranslated images. To further ensure the disentanglement between the views and other factors, we add adversarial training on the code. The results and ablation studies on MultiPIE and 3D chair datasets validate the effectiveness of the framework in cVAE and the designed module.

Yan Zhang, Zhilin Zheng, Binyu He et al.

Data uncertainty in practical person reID is ubiquitous, hence it requires not only learning the discriminative features, but also modeling the uncertainty based on the input. This paper proposes to learn the sample posterior and the class prior distribution in the latent space, so that not only representative features but also the uncertainty can be built by the model. The prior reflects the distribution of all data in the same class, and it is the trainable model parameters. While the posterior is the probability density of a single sample, so it is actually the feature defined on the input. We assume that both of them are in Gaussian form. To simultaneously model them, we put forward a distribution loss, which measures the KL divergence from the posterior to the priors in the manner of supervised learning. In addition, we assume that the posterior variance, which is essentially the uncertainty, is supposed to have the second-order characteristic. Therefore, a $Σ-$net is proposed to compute it by the high order representation from its input. Extensive experiments have been carried out on Market1501, DukeMTMC, MARS and noisy dataset as well.

Yan Zhang, Binyu He, Li Sun

Image features from a small local region often give strong evidence in person re-identification task. However, CNN suffers from paying too much attention on the most salient local areas, thus ignoring other discriminative clues, e.g., hair, shoes or logos on clothes. %BDB proposes to randomly drop one block in a batch to enlarge the high response areas. Although BDB has achieved remarkable results, there still room for improvement. In this work, we propose a Progressive Multi-stage feature Mix network (PMM), which enables the model to find out the more precise and diverse features in a progressive manner. Specifically, 1. to enforce the model to look for different clues in the image, we adopt a multi-stage classifier and expect that the model is able to focus on a complementary region in each stage. 2. we propose an Attentive feature Hard-Mix (A-Hard-Mix) to replace the salient feature blocks by the negative example in the current batch, whose label is different from the current sample. 3. extensive experiments have been carried out on reID datasets such as the Market-1501, DukeMTMC-reID and CUHK03, showing that the proposed method can boost the re-identification performance significantly.

Zhi Yan, Li Sun, Tom Duckett et al.

Human detection and tracking is an essential task for service robots, where the combined use of multiple sensors has potential advantages that are yet to be exploited. In this paper, we introduce a framework allowing a robot to learn a new 3D LiDAR-based human classifier from other sensors over time, taking advantage of a multisensor tracking system. The main innovation is the use of different detectors for existing sensors (i.e. RGB-D camera, 2D LiDAR) to train, online, a new 3D LiDAR-based human classifier, exploiting a so-called trajectory probability. Our framework uses this probability to check whether new detections belongs to a human trajectory, estimated by different sensors and/or detectors, and to learn a human classifier in a semi-supervised fashion. The framework has been implemented and tested on a real-world dataset collected by a mobile robot. We present experiments illustrating that our system is able to effectively learn from different sensors and from the environment, and that the performance of the 3D LiDAR-based human classification improves with the number of sensors/detectors used.

Li Sun, Cheng Zhao, Rustam Stolkin

This paper addresses the problem of RGBD object recognition in real-world applications, where large amounts of annotated training data are typically unavailable. To overcome this problem, we propose a novel, weakly-supervised learning architecture (DCNN-GPC) which combines parametric models (a pair of Deep Convolutional Neural Networks (DCNN) for RGB and D modalities) with non-parametric models (Gaussian Process Classification). Our system is initially trained using a small amount of labeled data, and then automatically prop- agates labels to large-scale unlabeled data. We first run 3D- based objectness detection on RGBD videos to acquire many unlabeled object proposals, and then employ DCNN-GPC to label them. As a result, our multi-modal DCNN can be trained end-to-end using only a small amount of human annotation. Finally, our 3D-based objectness detection and multi-modal DCNN are integrated into a real-time detection and recognition pipeline. In our approach, bounding-box annotations are not required and boundary-aware detection is achieved. We also propose a novel way to pretrain a DCNN for the depth modality, by training on virtual depth images projected from CAD models. We pretrain our multi-modal DCNN on public 3D datasets, achieving performance comparable to state-of-the-art methods on Washington RGBS Dataset. We then finetune the network by further training on a small amount of annotated data from our novel dataset of industrial objects (nuclear waste simulants). Our weakly supervised approach has demonstrated to be highly effective in solving a novel RGBD object recognition application which lacks of human annotations.