A number of recent self-supervised learning methods have shown impressive performance on image classification and other tasks. A somewhat bewildering variety of techniques have been used, not always with a clear understanding of the reasons for their benefits, especially when used in combination. Here we treat the embeddings of images as point particles and consider model optimization as a dynamic process on this system of particles. Our dynamic model combines an attractive force for similar images, a locally dispersive force to avoid local collapse, and a global dispersive force to achieve a globally-homogeneous distribution of particles. The dynamic perspective highlights the advantage of using a delayed-parameter image embedding (a la BYOL) together with multiple views of the same image. It also uses a purely-dynamic local dispersive force (Brownian motion) that shows improved performance over other methods and does not require knowledge of other particle coordinates. The method is called MSBReg which stands for (i) a Multiview centroid loss, which applies an attractive force to pull different image view embeddings toward their centroid, (ii) a Singular value loss, which pushes the particle system toward spatially homogeneous density, (iii) a Brownian diffusive loss. We evaluate downstream classification performance of MSBReg on ImageNet as well as transfer learning tasks including fine-grained classification, multi-class object classification, object detection, and instance segmentation. In addition, we also show that applying our regularization term to other methods further improves their performance and stabilize the training by preventing a mode collapse.
The assessment of general movements (GMs) in infants is a useful tool in the early diagnosis of neurodevelopmental disorders. However, its evaluation in clinical practice relies on visual inspection by experts, and an automated solution is eagerly awaited. Recently, video-based GMs classification has attracted attention, but this approach would be strongly affected by irrelevant information, such as background clutter in the video. Furthermore, for reliability, it is necessary to properly extract the spatiotemporal features of infants during GMs. In this study, we propose an automated GMs classification method, which consists of preprocessing networks that remove unnecessary background information from GMs videos and adjust the infant's body position, and a subsequent motion classification network based on a two-stream structure. The proposed method can efficiently extract the essential spatiotemporal features for GMs classification while preventing overfitting to irrelevant information for different recording environments. We validated the proposed method using videos obtained from 100 infants. The experimental results demonstrate that the proposed method outperforms several baseline models and the existing methods.
While fine-tuning pre-trained networks has become a popular way to train image segmentation models, such backbone networks for image segmentation are frequently pre-trained using image classification source datasets, e.g., ImageNet. Though image classification datasets could provide the backbone networks with rich visual features and discriminative ability, they are incapable of fully pre-training the target model (i.e., backbone+segmentation modules) in an end-to-end manner. The segmentation modules are left to random initialization in the fine-tuning process due to the lack of segmentation labels in classification datasets. In our work, we propose a method that leverages Pseudo Semantic Segmentation Labels (PSSL), to enable the end-to-end pre-training for image segmentation models based on classification datasets. PSSL was inspired by the observation that the explanation results of classification models, obtained through explanation algorithms such as CAM, SmoothGrad and LIME, would be close to the pixel clusters of visual objects. Specifically, PSSL is obtained for each image by interpreting the classification results and aggregating an ensemble of explanations queried from multiple classifiers to lower the bias caused by single models. With PSSL for every image of ImageNet, the proposed method leverages a weighted segmentation learning procedure to pre-train the segmentation network en masse. Experiment results show that, with ImageNet accompanied by PSSL as the source dataset, the proposed end-to-end pre-training strategy successfully boosts the performance of various segmentation models, i.e., PSPNet-ResNet50, DeepLabV3-ResNet50, and OCRNet-HRNetW18, on a number of segmentation tasks, such as CamVid, VOC-A, VOC-C, ADE20K, and CityScapes, with significant improvements. The source code is availabel at https://github.com/PaddlePaddle/PaddleSeg.
Time series are ubiquitous and therefore inherently hard to analyze and ultimately to label or cluster. With the rise of the Internet of Things (IoT) and its smart devices, data is collected in large amounts any given second. The collected data is rich in information, as one can detect accidents (e.g. cars) in real time, or assess injury/sickness over a given time span (e.g. health devices). Due to its chaotic nature and massive amounts of datapoints, timeseries are hard to label manually. Furthermore new classes within the data could emerge over time (contrary to e.g. handwritten digits), which would require relabeling the data. In this paper we present SuSL4TS, a deep generative Gaussian mixture model for semi-unsupervised learning, to classify time series data. With our approach we can alleviate manual labeling steps, since we can detect sparsely labeled classes (semi-supervised) and identify emerging classes hidden in the data (unsupervised). We demonstrate the efficacy of our approach with established time series classification datasets from different domains.
This paper presents the design of deep learning architectures which allow to classify the social relationship existing between two people who are walking in a side-by-side formation into four possible categories --colleagues, couple, family or friendship. The models are developed using Neural Networks or Recurrent Neural Networks to achieve the classification and are trained and evaluated using a database of readings obtained from humans performing an accompaniment process in an urban environment. The best achieved model accomplishes a relatively good accuracy in the classification problem and its results enhance partially the outcomes from a previous study [1]. Furthermore, the model proposed shows its future potential to improve its efficiency and to be implemented in a real robot.
We introduce the Few-Shot Object Learning (FewSOL) dataset for object recognition with a few images per object. We captured 336 real-world objects with 9 RGB-D images per object from different views. Object segmentation masks, object poses and object attributes are provided. In addition, synthetic images generated using 330 3D object models are used to augment the dataset. We investigated (i) few-shot object classification and (ii) joint object segmentation and few-shot classification with the state-of-the-art methods for few-shot learning and meta-learning using our dataset. The evaluation results show that there is still a large margin to be improved for few-shot object classification in robotic environments. Our dataset can be used to study a set of few-shot object recognition problems such as classification, detection and segmentation, shape reconstruction, pose estimation, keypoint correspondences and attribute recognition. The dataset and code are available at https://irvlutd.github.io/FewSOL.
In this study, we propose feature extraction for multimodal meme classification using Deep Learning approaches. A meme is usually a photo or video with text shared by the young generation on social media platforms that expresses a culturally relevant idea. Since they are an efficient way to express emotions and feelings, a good classifier that can classify the sentiment behind the meme is important. To make the learning process more efficient, reduce the likelihood of overfitting, and improve the generalizability of the model, one needs a good approach for joint feature extraction from all modalities. In this work, we proposed to use different multimodal neural network approaches for multimodal feature extraction and use the extracted features to train a classifier to identify the sentiment in a meme.
In this work, we present a multi-modal model for commercial product classification, that combines features extracted by multiple neural network models from textual (CamemBERT and FlauBERT) and visual data (SE-ResNeXt-50), using simple fusion techniques. The proposed method significantly outperformed the unimodal models' performance and the reported performance of similar models on our specific task. We did experiments with multiple fusing techniques and found, that the best performing technique to combine the individual embedding of the unimodal network is based on combining concatenation and averaging the feature vectors. Each modality complemented the shortcomings of the other modalities, demonstrating that increasing the number of modalities can be an effective method for improving the performance of multi-label and multimodal classification problems.
Binary classification with an imbalanced dataset is challenging. Models tend to consider all samples as belonging to the majority class. Although existing solutions such as sampling methods, cost-sensitive methods, and ensemble learning methods improve the poor accuracy of the minority class, these methods are limited by overfitting problems or cost parameters that are difficult to decide. We propose HADR, a hybrid approach with dimension reduction that consists of data block construction, dimentionality reduction, and ensemble learning with deep neural network classifiers. We evaluate the performance on eight imbalanced public datasets in terms of recall, G-mean, and AUC. The results show that our model outperforms state-of-the-art methods.
The use of artificial intelligence in the agricultural sector has been growing at a rapid rate to automate farming activities. Emergent farming technologies focus on mapping and classification of plants, fruits, diseases, and soil types. Although, assisted harvesting and pruning applications using deep learning algorithms are in the early development stages, there is a demand for solutions to automate such processes. This paper proposes the use of Deep Learning for the classification of trusses and runners of strawberry plants using semantic segmentation and dataset augmentation. The proposed approach is based on the use of noises (i.e. Gaussian, Speckle, Poisson and Salt-and-Pepper) to artificially augment the dataset and compensate the low number of data samples and increase the overall classification performance. The results are evaluated using mean average of precision, recall and F1 score. The proposed approach achieved 91%, 95% and 92% on precision, recall and F1 score, respectively, for truss detection using the ResNet101 with dataset augmentation utilising Salt-and-Pepper noise; and 83%, 53% and 65% on precision, recall and F1 score, respectively, for truss detection using the ResNet50 with dataset augmentation utilising Poisson noise.
Early detection of retinal diseases is one of the most important means of preventing partial or permanent blindness in patients. In this research, a novel multi-label classification system is proposed for the detection of multiple retinal diseases, using fundus images collected from a variety of sources. First, a new multi-label retinal disease dataset, the MuReD dataset, is constructed, using a number of publicly available datasets for fundus disease classification. Next, a sequence of post-processing steps is applied to ensure the quality of the image data and the range of diseases, present in the dataset. For the first time in fundus multi-label disease classification, a transformer-based model optimized through extensive experimentation is used for image analysis and decision making. Numerous experiments are performed to optimize the configuration of the proposed system. It is shown that the approach performs better than state-of-the-art works on the same task by 7.9% and 8.1% in terms of AUC score for disease detection and disease classification, respectively. The obtained results further support the potential applications of transformer-based architectures in the medical imaging field.
We apply methods of topological analysis to the attention graphs, calculated on the attention heads of the BERT model ( arXiv:1810.04805v2 ). Our research shows that the classifier built upon basic persistent topological features (namely, Betti numbers) of the trained neural network can achieve classification results on par with the conventional classification method. We show the relevance of such topological text representation on three text classification benchmarks. For the best of our knowledge, it is the first attempt to analyze the topology of an attention-based neural network, widely used for Natural Language Processing.
This paper presents our solution for the 2nd COVID-19 Competition, occurring in the framework of the AIMIA Workshop in the European Conference on Computer Vision (ECCV 2022). In our approach, we employ an effective 3D Contrastive Mixup Classification network for COVID-19 diagnosis on chest CT images, which is composed of contrastive representation learning and mixup classification. For the COVID-19 detection challenge, our approach reaches 0.9245 macro F1 score on 484 validation CT scans, which significantly outperforms the baseline method by 16.5%. In the COVID-19 severity detection challenge, our approach achieves 0.7186 macro F1 score on 61 validation samples, which also surpasses the baseline by 8.86%.
Video anomaly analysis is a core task actively pursued in the field of computer vision, with applications extending to real-world crime detection in surveillance footage. In this work, we address the task of human-related crime classification. In our proposed approach, the human body in video frames, represented as skeletal joints trajectories, is used as the main source of exploration. First, we introduce the significance of extending the ground truth labels for HR-Crime dataset and hence, propose a supervised and unsupervised methodology to generate trajectory-level ground truth labels. Next, given the availability of the trajectory-level ground truth, we introduce a trajectory-based crime classification framework. Ablation studies are conducted with various architectures and feature fusion strategies for the representation of the human trajectories. The conducted experiments demonstrate the feasibility of the task and pave the path for further research in the field.
Multi-label classification is a challenging task, particularly in domains where the number of labels to be predicted is large. Deep neural networks are often effective at multi-label classification of images and textual data. When dealing with tabular data, however, conventional machine learning algorithms, such as tree ensembles, appear to outperform competition. Random forest, being a popular ensemble algorithm, has found use in a wide range of real-world problems. Such problems include fraud detection in the financial domain, crime hotspot detection in the legal sector, and in the biomedical field, disease probability prediction when patient records are accessible. Since they have an impact on people's lives, these domains usually require decision-making systems to be explainable. Random Forest falls short on this property, especially when a large number of tree predictors are used. This issue was addressed in a recent research named LionForests, regarding single label classification and regression. In this work, we adapt this technique to multi-label classification problems, by employing three different strategies regarding the labels that the explanation covers. Finally, we provide a set of qualitative and quantitative experiments to assess the efficacy of this approach.
Deep learning-based classification of rare anemia disorders is challenged by the lack of training data and instance-level annotations. Multiple Instance Learning (MIL) has shown to be an effective solution, yet it suffers from low accuracy and limited explainability. Although the inclusion of attention mechanisms has addressed these issues, their effectiveness highly depends on the amount and diversity of cells in the training samples. Consequently, the poor machine learning performance on rare anemia disorder classification from blood samples remains unresolved. In this paper, we propose an interpretable pooling method for MIL to address these limitations. By benefiting from instance-level information of negative bags (i.e., homogeneous benign cells from healthy individuals), our approach increases the contribution of anomalous instances. We show that our strategy outperforms standard MIL classification algorithms and provides a meaningful explanation behind its decisions. Moreover, it can denote anomalous instances of rare blood diseases that are not seen during the training phase.
The Brazilian Supreme Court receives tens of thousands of cases each semester. Court employees spend thousands of hours to execute the initial analysis and classification of those cases -- which takes effort away from posterior, more complex stages of the case management workflow. In this paper, we explore multimodal classification of documents from Brazil's Supreme Court. We train and evaluate our methods on a novel multimodal dataset of 6,510 lawsuits (339,478 pages) with manual annotation assigning each page to one of six classes. Each lawsuit is an ordered sequence of pages, which are stored both as an image and as a corresponding text extracted through optical character recognition. We first train two unimodal classifiers: a ResNet pre-trained on ImageNet is fine-tuned on the images, and a convolutional network with filters of multiple kernel sizes is trained from scratch on document texts. We use them as extractors of visual and textual features, which are then combined through our proposed Fusion Module. Our Fusion Module can handle missing textual or visual input by using learned embeddings for missing data. Moreover, we experiment with bi-directional Long Short-Term Memory (biLSTM) networks and linear-chain conditional random fields to model the sequential nature of the pages. The multimodal approaches outperform both textual and visual classifiers, especially when leveraging the sequential nature of the pages.
Thyroid nodule classification aims at determining whether the nodule is benign or malignant based on a given ultrasound image. However, the label obtained by the cytological biopsy which is the golden standard in clinical medicine is not always consistent with the ultrasound imaging TI-RADS criteria. The information difference between the two causes the existing deep learning-based classification methods to be indecisive. To solve the Inconsistent Label problem, we propose an Adaptive Curriculum Learning (ACL) framework, which adaptively discovers and discards the samples with inconsistent labels. Specifically, ACL takes both hard sample and model certainty into account, and could accurately determine the threshold to distinguish the samples with Inconsistent Label. Moreover, we contribute TNCD: a Thyroid Nodule Classification Dataset to facilitate future related research on the thyroid nodules. Extensive experimental results on TNCD based on three different backbone networks not only demonstrate the superiority of our method but also prove that the less-is-more principle which strategically discards the samples with Inconsistent Label could yield performance gains. Source code and data are available at https://github.com/chenghui-666/ACL/.
Class distribution plays an important role in learning deep classifiers. When the proportion of each class in the test set differs from the training set, the performance of classification nets usually degrades. Such a label distribution shift problem is common in medical diagnosis since the prevalence of disease vary over location and time. In this paper, we propose the first method to tackle label shift for medical image classification, which effectively adapt the model learned from a single training label distribution to arbitrary unknown test label distribution. Our approach innovates distribution calibration to learn multiple representative classifiers, which are capable of handling different one-dominating-class distributions. When given a test image, the diverse classifiers are dynamically aggregated via the consistency-driven test-time adaptation, to deal with the unknown test label distribution. We validate our method on two important medical image classification tasks including liver fibrosis staging and COVID-19 severity prediction. Our experiments clearly show the decreased model performance under label shift. With our method, model performance significantly improves on all the test datasets with different label shifts for both medical image diagnosis tasks.
Intestinal parasitic infections, as a leading causes of morbidity worldwide, still lacks time-saving, high-sensitivity and user-friendly examination method. The development of deep learning technique reveals its broad application potential in biological image. In this paper, we apply several object detectors such as YOLOv5 and variant cascadeRCNNs to automatically discriminate parasitic eggs in microscope images. Through specially-designed optimization including raw data augmentation, model ensemble, transfer learning and test time augmentation, our model achieves excellent performance on challenge dataset. In addition, our model trained with added noise gains a high robustness against polluted input, which further broaden its applicability in practice.