Manganese self-boosting hollow nanoenzymes with glutathione depletion for synergistic cancer chemo-chemodynamic therapy.

Chemodynamic therapy (CDT) has outstanding potential as a combination therapy to treat cancer. However, the effectiveness of CDT in the treatment of solid tumors is limited by the overexpression of glutathione (GSH) in the tumor microenvironment (TME). GSH overexpression diminishes oxidative stress and attenuates chemotherapeutic drug-induced apoptosis in cancer cells. To counter these effects, a synergistic CDT/chemotherapy cancer treatment, involving the use of a multifunctional bioreactor of hollow manganese dioxide (HMnO2) loaded with cisplatin (CDDP), was developed. Metal nanoenzymes that can auto-degrade to produce Mn2+ exhibit Fenton-like, GSH-peroxidase-like activity, which effectively depletes GSH in the TME to attenuate the tumor antioxidant capacity. In an acidic environment, Mn2+ catalyzed the decomposition of intra-tumor H2O2 into highly toxic ·OH as a CDT. HMnO2 with large pores, pore volume, and surface area exhibited a high CDDP loading capacity (>0.6 g-1). Treatment with CDDP-loaded HMnO2 increased the intratumor Pt-DNA content, leading to the up-regulation of γ-H2Aχ and an increase in tumor tissue damage. The decreased GSH triggered by HMnO2 auto-degradation protected Mn2+-generated ·OH from scavenging to amplify oxidative stress and enhance the efficacy of CDT. The nanoenzymes with encapsulated chemotherapeutic agents deplete GSH and remodel the TME. Thus, tumor CDT/chemotherapy combination therapy is an effective therapeutic strategy.

On Impact Damage and Repair of Composite Honeycomb Sandwich Structures.

This study is conducted on glass fiber-reinforced composite honeycomb sandwich structures by introducing delamination damage through low-velocity impact tests, establishing a three-dimensional progressive damage analysis model, and evaluating the delamination damage characteristics and laws of honeycomb sandwich structures under different impact energies through experiments. Repair techniques and process parameters for delamination damage are explored. It is found that as the impact energy increases, the damage area of honeycomb sandwich panels also increases, and the delamination damage extends from the impact center to the surrounding areas, accompanied by damage such as fiber fracture and matrix cracking. The strength recovery rates of sandwich panels at impact energies of 5 J, 15 J, and 25 J after repair are 71.90%, 65.89%, and 67.10%, respectively, which has a considerable repair effect. In addition, a progressive damage model for low-velocity impact on the composite honeycomb sandwich structure is established, and its accuracy and reliability are verified.

OBMO: One Bounding Box Multiple Objects for Monocular 3D Object Detection.

Compared to typical multi-sensor systems, monocular 3D object detection has attracted much attention due to its simple configuration. However, there is still a significant gap between LiDAR-based and monocular-based methods. In this paper, we find that the ill-posed nature of monocular imagery can lead to depth ambiguity. Specifically, objects with different depths can appear with the same bounding boxes and similar visual features in the 2D image. Unfortunately, the network cannot accurately distinguish different depths from such non-discriminative visual features, resulting in unstable depth training. To facilitate depth learning, we propose a simple yet effective plug-and-play module, One Bounding Box Multiple Objects (OBMO). Concretely, we add a set of suitable pseudo labels by shifting the 3D bounding box along the viewing frustum. To constrain the pseudo-3D labels to be reasonable, we carefully design two label scoring strategies to represent their quality. In contrast to the original hard depth labels, such soft pseudo labels with quality scores allow the network to learn a reasonable depth range, boosting training stability and thus improving final performance. Extensive experiments on KITTI and Waymo benchmarks show that our method significantly improves state-of-the-art monocular 3D detectors by a significant margin (The improvements under the moderate setting on KITTI validation set are 1.82 ~ 10.91% mAP in BEV and 1.18 ~ 9.36% mAP in 3D). Codes have been released at https://github.com/mrsempress/OBMO.

Design and Analysis of a New Multi-Part Composite Frangible Cover.

In this paper, a new multi-part composite frangible cover (MCFC) was designed and fabricated. The frangible cover, manufactured with a traditional manual lay-up method, is designed to conduct a simulated missile launch test using a specially developed test device. A weak zone structure of the composite multi-part frangible cover was designed, and the separation process of the cover was studied by numerical simulation. Based on the strength envelope of the weak zone and the equal-strength design principle, a design method for the weak zone structure of the composite multi-part frangible cover was proposed. A finite element model of the composite multi-part frangible cover was established, and the separation process was numerically simulated and analyzed. Afterward, the verification experiments were carried out. Close agreements between the numerical and experimental results are observed.

X-view: Non-egocentric Multi-View 3D Object Detector.

3D object detection algorithms for autonomous driving reason about 3D obstacles either from 3D birds-eye view or perspective view or both. Recent works attempt to improve the detection performance via mining and fusing from multiple egocentric views. Although the egocentric perspective view alleviates some weaknesses of the birds-eye view, the sectored grid partition becomes so coarse in the distance that the targets and surrounding context mix together, which makes the features less discriminative. In this paper, we generalize the research on 3D multi-view learning and propose a novel multi-view-based 3D detection method, named X-view, to overcome the drawbacks of the multi-view methods. Specifically, X-view breaks through the traditional limitation about the perspective view whose original point must be consistent with the 3D Cartesian coordinate. X-view is designed as a general paradigm that can be applied on almost any 3D detectors based on LiDAR with only little increment of running time, no matter it is voxel/grid-based or raw-point-based. We conduct experiments on KITTI [1] and NuScenes [2] datasets to demonstrate the robustness and effectiveness of our proposed X-view. The results show that X-view obtains consistent improvements when combined with mainstream state-of-the-art 3D methods.

Investigation on Musculoskeletal Injury and Psychological Empowerment of Reflexologists in Taiwan: Analysis of the Recognition to Alternative Therapy.

Many studies have proven that reflexology has been used as a complementary medical treatment. Therefore, the government has started to plan an examination system for reflexology personnel to ensure the quality of service. Reflexologists work long hours, have heavy workloads, and perform poses that do not conform to human factors, which often cause musculoskeletal fatigue. The purpose of this study is to understand the musculoskeletal pain conditions of reflexologists, the psychological empowerment status, and the perceptions of complementary medicine therapy. The data for this study were obtained in two ways: (1) 59 practitioners were surveyed by using a face-to-face questionnaire and (2) a semi-structured interview was carried out for 10 practitioners. This study discovered the following: (1) Reflexology practitioners have musculoskeletal discomfort symptoms in body parts, including the left shoulder (25.4%), left hand or wrist (25.4%), lower back (25.4%), right shoulder (23.7%), left elbow or forearm (22%). (2) Reflexology practitioners are highly psychologically empowered to work. (3) The practitioners of foot therapy hold a positive attitude towards foot therapy and believe that foot therapy is a natural therapy, which is self-serving and can help others. (4) Most reflexologists support the government's desire to promote the reflexology examination system and are willing to help develop the policy. (5) The height of most reflexologist work chairs does not match the height of the guest's seat and is not ergonomic.

The Influence of the COVID-19 Pandemic on Emergency Medical Services to Out-of-Hospital Cardiac Arrests in a Low-Incidence Urban City: An Observational Epidemiological Analysis.

The Emergency Medical Services (EMS) system faced overwhelming challenges during the coronavirus disease 2019 (COVID-19) pandemic. However, further information is required to determine how the pandemic affected the EMS response and the clinical outcomes of out-of-hospital cardiac arrest (OHCA) patients in COVID-19 low-incidence cities. A retrospective study was conducted in Chiayi, Taiwan, a COVID-19 low-incidence urban city. We compared the outcomes and rescue records before (2018-2019) and during (2020-2021) the COVID-19 pandemic. A total of 567 patients before and 497 during the pandemic were enrolled. Multivariate analysis revealed that the COVID-19 pandemic had no significant influence on the achievement of return of spontaneous circulation (ROSC) and sustained ROSC but was associated with lower probabilities of survival to discharge (aOR = 0.43, 95% CI: 0.21-0.89, p = 0.002) and discharge with favorable neurologic outcome among OHCA patients (aOR = 0.35, 95% CI: 0.16-0.77, p = 0.009). Patients' ages and OHCA locations were also discovered to be independently related to survival results. The overall impact of longer EMS rescue times on survival outcomes during the pandemic was not significant, with an exception of the specific group that experienced prolonged rescue times (total EMS time > 21 min).

Toward Better Accuracy-Efficiency Trade-Offs: Divide and Co-Training.

The width of a neural network matters since increasing the width will necessarily increase the model capacity. However, the performance of a network does not improve linearly with the width and soon gets saturated. In this case, we argue that increasing the number of networks (ensemble) can achieve better accuracy-efficiency trade-offs than purely increasing the width. To prove it, one large network is divided into several small ones regarding its parameters and regularization components. Each of these small networks has a fraction of the original one's parameters. We then train these small networks together and make them see various views of the same data to increase their diversity. During this co-training process, networks can also learn from each other. As a result, small networks can achieve better ensemble performance than the large one with few or no extra parameters or FLOPs, i. e., achieving better accuracy-efficiency trade-offs. Small networks can also achieve faster inference speed than the large one by concurrent running. All of the above shows that the number of networks is a new dimension of model scaling. We validate our argument with 8 different neural architectures on common benchmarks through extensive experiments.

Effect of Interlaminar Toughness on the Residual Compressive Capacity of Carbon Fiber Laminates with Different Types of Delamination.

In this paper, the effect of interlaminar properties and the type of delamination defects on the residual compression properties of carbon fiber laminates were experimentally investigated. A new method, which employed magnetic force to guide the arrangement direction of stainless steel particles between layers of laminates, was adopted to improve the interlayer toughness. The digital image correlation, C-scan, and micro-CT were used to measure and identify the compression failure damages. Test results showed that the compressive strength of the intact carbon fiber laminates was 299.37 MPa, and the one of specimens containing the deeply buried delamination, the through-width delamination, and the surface delamination decreased by 55.98 MPa, 58.69 MPa, and 60.23 MPa, respectively. The compressive strength of the specimens containing the deeply buried delamination only decreased by 14.01 MPa when the mode I toughness increased by 81.88%, and the specimen containing the surface delamination only decreased by 30.86 MPa when the mode II fracture toughness increased by 87.72%. However, improving the fracture toughness could not strengthen the specimens containing the through-width delamination. Moreover, a qualitative dynamic damage relationship, which described the relationship between delamination expansion and compression damage vividly, was proposed. The reason the increase of the toughness could improve the residual compression performance of the laminates containing delamination was that the higher fracture toughness hindered the secondary expansion of the delamination during the compression process so that the delamination area could almost remain unchanged.

Preoperative Changes of Lung Nodule on Computed Tomography and Their Relationship With Pathological Outcomes.

Background: Whether changes of lung nodules on computed tomography could bring us helpful information related to their pathological outcomes remained unclear. Materials and Methods: This retrospective study was carried out among 1,185 cases of lung nodules in Shanghai Chest Hospital from January 2015 to April 2017, which did not shrink or disappear after preoperative follow-up over three months. Their imaging features, changes, and clinical characteristics were collected. A separate analysis was performed in nodules with or without growth in long-axis diameter after follow-up, searching significant changes related to nodule malignancy and the median interval of follow-up for reference. Further study was performed similarly in malignant nodules for discrimination of malignant grading. Results: Most nodules were stable (n = 885, 75%), whereas others grew (n = 300, 25%). For predicting nodule malignancy, increase in density (>10 Hounsfield units, median follow-up of 549 days) played an important role in growing group whereas it failed in stable group, and the increase in size was less significant in growing group. For discrimination of malignant grading, increase in density (>70 Hounsfield units, median follow-up of 366 days) showed its significance in stable group, and so did increase in size in growing group (maximum diameter growth >3.3 mm, median follow-up of 549 days, or average diameter growth >3.1 mm, median follow-up of 625 days). Conclusions: There were significant changes of lung nodules by follow-up on computed tomography, related to their pathological outcomes. The predictive power of increase in density or size varied in different situations, whereas all referred to a long-time preoperative follow-up.

Progressive Transfer Learning.

Model fine-tuning is a widely used transfer learning approach in person Re-identification (ReID) applications, which fine-tuning a pre-trained feature extraction model into the target scenario instead of training a model from scratch. It is challenging due to the significant variations inside the target scenario, e.g., different camera viewpoint, illumination changes, and occlusion. These variations result in a gap between each mini-batch's distribution and the whole dataset's distribution when using mini-batch training. In this paper, we study model fine-tuning from the perspective of the aggregation and utilization of the dataset's global information when using mini-batch training. Specifically, we introduce a novel network structure called Batch-related Convolutional Cell (BConv-Cell), which progressively collects the dataset's global information into a latent state and uses it to rectify the extracted feature. Based on BConv-Cells, we further proposed the Progressive Transfer Learning (PTL) method to facilitate the model fine-tuning process by jointly optimizing BConv-Cells and the pre-trained ReID model. Empirical experiments show that our proposal can greatly improve the ReID model's performance on MSMT17, Market-1501, CUHK03, and DukeMTMC-reID datasets. Moreover, we extend our proposal to the general image classification task. The experiments in several image classification benchmark datasets demonstrate that our proposal can significantly improve baseline models' performance. The code has been released at https://github.com/ZJULearning/PTL.

High Dimensional Similarity Search With Satellite System Graph: Efficiency, Scalability, and Unindexed Query Compatibility.

Approximate nearest neighbor search (ANNS) in high-dimensional space is essential in database and information retrieval. Recently, there has been a surge of interest in exploring efficient graph-based indices for the ANNS problem. Among them, navigating spreading-out graph (NSG) provides fine theoretical analysis and achieves state-of-the-art performance. However, we find there are several limitations with NSG: 1) NSG has no theoretical guarantee on nearest neighbor search when the query is not indexed in the database; and 2) NSG is too sparse which harms the search performance. In addition, NSG suffers from high indexing complexity. To address above problems, we propose the satellite system graphs (SSG) and a practical variant NSSG. Specifically, we propose a novel pruning strategy to produce SSGs from the complete graph. SSGs define a new family of MSNETs in which the out-edges of each node are distributed evenly in all directions. Each node in the graph builds effective connections to its neighborhood omnidirectionally, whereupon we derive SSG's excellent theoretical properties for both indexed and unindexed queries. We can adaptively adjust the sparsity of an SSG with a hyper-parameter to optimize the search performance. Further, NSSG is proposed to reduce the indexing complexity of the SSG for large-scale applications. Both theoretical and extensive experimental analysis are provided to demonstrate the strengths of the proposed approach over the existing representative algorithms. Our code has been released at https://github.com/ZJULearning/SSG.

Foot Reflexology: Recent Research Trends and Prospects.

Foot reflexology is a non-invasive complementary therapy that is increasingly being accepted by modern people in recent years. To understand the research trends and prospects of foot reflexology in the past 31 years, this study used the Web of Science core collection as the data source and two visualization tools, COOC and VOSviewer, to analyze the literature related to the field of foot reflexology from 1991 to 2021. This study found that the number of articles published in the field of foot reflexology has been increasing year by year, and the top three journals with the most articles are Complementary Therapies in Clinical Practice, Therapies in Medicine, and the Journal of Alternative and Complementary Medicine. The top three most prolific authors are Wyatt, Sikorskii, and Victorson, and the core institutions in the field of foot reflexology are Michigan State University, Northwestern University, Tehran University of Medical Sciences, and the University of Exeter. Foot reflexology has been shown to have a moderating effect on anxiety, fatigue, and cancer, and is a topic of ongoing and future research. This study uses this bibliometric analysis of foot reflexology literature to provide an overview of prior knowledge and a reference direction for modern preventive medicine.

Single-Cell Microwell Platform Reveals Circulating Neural Cells as a Clinical Indicator for Patients with Blood-Brain Barrier Breakdown.

Central nervous system diseases commonly occur with the destruction of the blood-brain barrier. As a primary cause of morbidity and mortality, stroke remains unpredictable and lacks cellular biomarkers that accurately quantify its occurrence and development. Here, we identify NeuN+/CD45-/DAPI+ phenotype nonblood cells in the peripheral blood of mice subjected to middle cerebral artery occlusion (MCAO) and stroke patients. Since NeuN is a specific marker of neural cells, we term these newly identified cells as circulating neural cells (CNCs). We find that the enumeration of CNCs in the blood is significantly associated with the severity of brain damage in MCAO mice (p < 0.05). Meanwhile, the number of CNCs is significantly higher in stroke patients than in negative subjects (p < 0.0001). These findings suggest that the amount of CNCs in circulation may serve as a clinical indicator for the real-time prognosis and progression monitor of the occurrence and development of ischemic stroke and other nervous system disease.

MALAT1 rs619586 A/G polymorphisms are associated with decreased risk of lung cancer.

ABSTRACT: Lung cancer is the leading cause of cancer-associated mortality worldwide. Genetic factors are reported to play important roles in lung carcinogenesis. To evaluate genetic susceptibility, we conducted a hospital-based case-control study on the effects of functional single nucleotide polymorphisms (SNPs) in long non-coding RNAs (lncRNAs) and microRNAs on lung cancer development. A total of 917 lung cancer cases and 925 control subjects were recruited. The MALAT1 rs619586 A/G genotype frequencies between patient and control groups were significantly different (P < .001), specifically, 83.85% vs 75.88% (AA), 15.60% vs 21.79% (AG), and 0.55% vs 2.32% (GG). When the homozygous genotype MALAT1 rs619586 AA was used as the reference group, AG (AG vs AA: adjusted odds ratio [OR] 0.65, 95% confidential interval [CI] 0.51-0.83, P = .001) and GG genotypes were associated with significantly decreased risk of lung cancer (GG vs AA: adjusted OR 0.22, 95% CI 0.08-0.59, P = .003). In the dominant model, MALAT1 rs619586 AG/GG variants were also associated with a significantly decreased risk of lung cancer (adjusted OR 0.61, 95% CI 0.48-0.78, P < .001). In the recessive model, when MALAT1 rs619586 AA/AG genotypes were used as the reference group, the GG homozygous genotype was also associated with significantly decreased risk for lung cancer (adjusted OR 0.24, 95% CI 0.09-0.64, P = .004). Hsa-miR-34b/c rs4938723 T > C, pri-miR-124-1 rs531564 C > G and hsa-miR-423 rs6505162 C > A SNPs were not associated with lung cancer risk. Our collective data indicated that MALAT1 rs619586 A/G SNPs significantly reduced the risk of lung cancer. Large-scale studies on different ethnic populations and tissue-specific biological characterization are required to validate the current findings.

Preoperative peripheral blood neutrophil-to-lymphocyte ratios (NLR) and platelet-to-lymphocyte ratio (PLR) related nomograms predict the survival of patients with limited-stage small-cell lung cancer.

BACKGROUND: We aim to establish neutrophil-to-lymphocyte ratio (NLR) and platelet-to-lymphocyte ratio (PLR) related nomograms based on the clinical data and peripheral blood markers to predict the survivals of patients with limited-stage small-cell lung cancer (LS-SCLC). METHODS: A total of 299 LS-SCLC patients after surgery were enrolled in this study. Univariate and multivariate analyses were conducted to select independent prognostic factors to develop the nomograms and then subjected to bootstrap internal validation. The optimal cutoff value of NLR and PLR before surgery was calculated by X-tile (version 3.6.1) and the overall survival (OS) was analyzed by Kaplan-Meier method and compared by log-rank test. RESULTS: According to the X-tile calculation, the NLR value and PLR cutoff values are 2.6 and 156.7, respectively. The prognosis of patients with elevated NLR or PLR value was significantly worse than patients with lower NLR (HR =1.798, 95% CI: 1.284-2.518, P=0.001) or PLR (HR =1.781, 95% CI: 1.318-2.407, P<0.001) value. Two Nomograms were developed according to the two multivariate cox regression models based on NLR and PLR. Concordance index (C-index) curves and calibration curves show that the two models have a better effect in predicting prognosis. At the same time, compared with the tumor node metastasis (TNM) staging system, our models also show better accuracy and stability. CONCLUSIONS: Elevated NLR and PLR predict poor prognosis in their respective nomograms in patients with LS-SCLC.

Complementary Pseudo Labels for Unsupervised Domain Adaptation On Person Re-Identification.

In recent years, supervised person re-identification (re-ID) models have received increasing studies. However, these models trained on the source domain always suffer dramatic performance drop when tested on an unseen domain. Existing methods are primary to use pseudo labels to alleviate this problem. One of the most successful approaches predicts neighbors of each unlabeled image and then uses them to train the model. Although the predicted neighbors are credible, they always miss some hard positive samples, which may hinder the model from discovering important discriminative information of the unlabeled domain. In this paper, to complement these low recall neighbor pseudo labels, we propose a joint learning framework to learn better feature embeddings via high precision neighbor pseudo labels and high recall group pseudo labels. The group pseudo labels are generated by transitively merging neighbors of different samples into a group to achieve higher recall. However, the merging operation may cause subgroups in the group due to imperfect neighbor predictions. To utilize these group pseudo labels properly, we propose using a similarity-aggregating loss to mitigate the influence of these subgroups by pulling the input sample towards the most similar embeddings. Extensive experiments on three large-scale datasets demonstrate that our method can achieve state-of-the-art performance under the unsupervised domain adaptation re-ID setting.

ES-Net: Erasing Salient Parts to Learn More in Re-Identification.

As an instance-level recognition problem, re-identification (re-ID) requires models to capture diverse features. However, with continuous training, re-ID models pay more and more attention to the salient areas. As a result, the model may only focus on few small regions with salient representations and ignore other important information. This phenomenon leads to inferior performance, especially when models are evaluated on small inter-identity variation data. In this paper, we propose a novel network, Erasing-Salient Net (ES-Net), to learn comprehensive features by erasing the salient areas in an image. ES-Net proposes a novel method to locate the salient areas by the confidence of objects and erases them efficiently in a training batch. Meanwhile, to mitigate the over-erasing problem, this paper uses a trainable pooling layer P-pooling that generalizes global max and global average pooling. Experiments are conducted on two specific re-identification tasks (i.e., Person re-ID, Vehicle re-ID). Our ES-Net outperforms state-of-the-art methods on three Person re-ID benchmarks and two Vehicle re-ID benchmarks. Specifically, mAP / Rank-1 rate: 88.6% / 95.7% on Market1501, 78.8% / 89.2% on DuckMTMC-reID, 57.3% / 80.9% on MSMT17, 81.9% / 97.0% on Veri-776, respectively. Rank-1 / Rank-5 rate: 83.6% / 96.9% on VehicleID (Small), 79.9% / 93.5% on VehicleID (Medium), 76.9% / 90.7% on VehicleID (Large), respectively. Moreover, the visualized salient areas show human-interpretable visual explanations for the ranking results.

Query-biased Self-attentive Network for Query-focused Video Summarization.

This paper addresses the task of query-focused video summarization, which takes user queries and long videos as inputs and generates query-focused video summaries. Compared to video summarization, which mainly concentrates on finding the most diverse and representative visual contents as a summary, the task of query-focused video summarization considers the user's intent and the semantic meaning of generated summary. In this paper, we propose a method, named query-biased self-attentive network (QSAN) to tackle this challenge. Our key idea is to utilize the semantic information from video descriptions to generate a generic summary and then to combine the information from the query to generate a query-focused summary. Specifically, we first propose a hierarchical self-attentive network to model the relative relationship at three levels, which are different frames from a segment, different segments of the same video, textual information of video description and its related visual contents. We train the model on video caption dataset and employ a reinforced caption generator to generate a video description, which can help us locate important frames or shots. Then we build a query-aware scoring module to compute the query-relevant score for each shot and generate the query-focused summary. Extensive experiments on the benchmark dataset demonstrate the competitive performance of our approach compared to some methods.

SIF: Self-Inspirited Feature Learning for Person Re-identification.

The re-identification (ReID) task has received increasing studies in recent years and its performance has gained significant improvement. The progress mainly comes from searching for new network structures to learn person representations. Most of these networks are trained using the classic stochastic gradient descent optimizer. However, limited efforts have been made to explore potential performance of existing ReID networks directly by better training scheme, which leaves a large space for ReID research. In this paper, we propose a Self-Inspirited Feature Learning (SIF) method to enhance performance of given ReID networks from the viewpoint of optimization. We design a simple adversarial learning scheme to encourage a network to learn more discriminative person representation. In our method, an auxiliary branch is added into the network only in the training stage, while the structure of the original network stays unchanged during the testing stage. In summary, SIF has three aspects of advantages: (1) it is designed under general setting; (2) it is compatible with many existing feature learning networks on the ReID task; (3) it is easy to implement and has steady performance. We evaluate the performance of SIF on three public ReID datasets: Market1501, DuckMTMC-reID, and CUHK03(both labeled and detected). The results demonstrate significant improvement in performance brought by SIF. We also apply SIF to obtain state-of-the-art results on all the three datasets. Specifically, mAP / Rank-1 accuracy are: 87.6% / 95.2% (without re-rank) on Market1501, 79.4% / 89.8% on DuckMTMC-reID, 77.0% / 79.5% on CUHK03 (labeled) and 73.9% / 76.6% on CUHK03 (detected), respectively. The code of SIF will be available soon.