Several recent publications have proposed methods for mapping images into continuous semantic embedding spaces. In some cases the embedding space is trained jointly with the image transformation. In other cases the semantic embedding space is established by an independent natural language processing task, and then the image transformation into that space is learned in a second stage. Proponents of these image embedding systems have stressed their advantages over the traditional \nway{} classification framing of image understanding, particularly in terms of the promise for zero-shot learning -- the ability to correctly annotate images of previously unseen object categories. In this paper, we propose a simple method for constructing an image embedding system from any existing \nway{} image classifier and a semantic word embedding model, which contains the $\n$ class labels in its vocabulary. Our method maps images into the semantic embedding space via convex combination of the class label embedding vectors, and requires no additional training. We show that this simple and direct method confers many of the advantages associated with more complex image embedding schemes, and indeed outperforms state of the art methods on the ImageNet zero-shot learning task.
Multi-sensor fusion is widely used in the environment perception system of the autonomous vehicle. It solves the interference caused by environmental changes and makes the whole driving system safer and more reliable. In this paper, a novel visible and near-infrared fusion method based on texture information is proposed to enhance unstructured environmental images. It aims at the problems of artifact, information loss and noise in traditional visible and near infrared image fusion methods. Firstly, the structure information of the visible image (RGB) and the near infrared image (NIR) after texture removal is obtained by relative total variation (RTV) calculation as the base layer of the fused image; secondly, a Bayesian classification model is established to calculate the noise weight and the noise information and the noise information in the visible image is adaptively filtered by joint bilateral filter; finally, the fused image is acquired by color space conversion. The experimental results demonstrate that the proposed algorithm can preserve the spectral characteristics and the unique information of visible and near-infrared images without artifacts and color distortion, and has good robustness as well as preserving the unique texture.
Restricted Boltzmann Machine (RBM) is an energy based, undirected graphical model. It is commonly used for unsupervised and supervised machine learning. Typically, RBM is trained using contrastive divergence (CD). However, training with CD is slow and does not estimate exact gradient of log-likelihood cost function. In this work, the model expectation of gradient learning for RBM has been calculated using a quantum annealer (D-Wave 2000Q), which is much faster than Markov chain Monte Carlo (MCMC) used in CD. Training and classification results are compared with CD. The classification accuracy results indicate similar performance of both methods. Image reconstruction as well as log-likelihood calculations are used to compare the performance of quantum and classical algorithms for RBM training. It is shown that the samples obtained from quantum annealer can be used to train a RBM on a 64-bit `bars and stripes' data set with classification performance similar to a RBM trained with CD. Though training based on CD showed improved learning performance, training using a quantum annealer eliminates computationally expensive MCMC steps of CD.
Annotating medical images for disease detection is often tedious and expensive. Moreover, the available training samples for a given task are generally scarce and imbalanced. These conditions are not conducive for learning effective deep neural models. Hence, it is common to 'transfer' neural networks trained on natural images to the medical image domain. However, this paradigm lacks in performance due to the large domain gap between the natural and medical image data. To address that, we propose a novel concept of Pre-text Representation Transfer (PRT). In contrast to the conventional transfer learning, which fine-tunes a source model after replacing its classification layers, PRT retains the original classification layers and updates the representation layers through an unsupervised pre-text task. The task is performed with (original, not synthetic) medical images, without utilizing any annotations. This enables representation transfer with a large amount of training data. This high-fidelity representation transfer allows us to use the resulting model as a more effective feature extractor. Moreover, we can also subsequently perform the traditional transfer learning with this model. We devise a collaborative representation based classification layer for the case when we leverage the model as a feature extractor. We fuse the output of this layer with the predictions of a model induced with the traditional transfer learning performed over our pre-text transferred model. The utility of our technique for limited and imbalanced data classification problem is demonstrated with an extensive five-fold evaluation for three large-scale models, tested for five different class-imbalance ratios for CT based COVID-19 detection. Our results show a consistent gain over the conventional transfer learning with the proposed method.
Convolutional Neural Networks (CNNs) have achieved impressive performance on many computer vision related tasks, such as object detection, image recognition, image retrieval, etc. These achievements benefit from the CNNs' outstanding capability to learn discriminative features with deep layers of neuron structures and iterative training process. This has inspired the EEG research community to adopt CNN in performing EEG classification tasks. However, CNNs learned features are not immediately interpretable, causing a lack of understanding of the CNNs' internal working mechanism. To improve CNN interpretability, CNN visualization methods are applied to translate the internal features into visually perceptible patterns for qualitative analysis of CNN layers. Many CNN visualization methods have been proposed in the Computer Vision literature to interpret the CNN network structure, operation, and semantic concept, yet applications to EEG data analysis have been limited. In this work we use 3 different methods to extract EEG-relevant features from a CNN trained on raw EEG data: optimal samples for each classification category, activation maximization, and reverse convolution. We applied these methods to a high-performing Deep Learning model with state-of-the-art performance for an EEG sex classification task, and show that the model features a difference in the theta frequency band. We show that visualization of a CNN model can reveal interesting EEG results. Using these tools, EEG researchers using Deep Learning can better identify the learned EEG features, possibly identifying new class relevant biomarkers.
Vision-language models have showcased impressive zero-shot classification capabilities when equipped with suitable text prompts. Previous studies have shown the effectiveness of test-time prompt tuning; however, these methods typically require per-image prompt adaptation during inference, which incurs high computational budgets and limits scalability and practical deployment. To overcome this issue, we introduce Self-TPT, a novel framework leveraging Self-supervised learning for efficient Test-time Prompt Tuning. The key aspect of Self-TPT is that it turns to efficient predefined class adaptation via self-supervised learning, thus avoiding computation-heavy per-image adaptation at inference. Self-TPT begins by co-training the self-supervised and the classification task using source data, then applies the self-supervised task exclusively for test-time new class adaptation. Specifically, we propose Contrastive Prompt Learning (CPT) as the key task for self-supervision. CPT is designed to minimize the intra-class distances while enhancing inter-class distinguishability via contrastive learning. Furthermore, empirical evidence suggests that CPT could closely mimic back-propagated gradients of the classification task, offering a plausible explanation for its effectiveness. Motivated by this finding, we further introduce a gradient matching loss to explicitly enhance the gradient similarity. We evaluated Self-TPT across three challenging zero-shot benchmarks. The results consistently demonstrate that Self-TPT not only significantly reduces inference costs but also achieves state-of-the-art performance, effectively balancing the efficiency-efficacy trade-off.
Deep learning has revolutionized the field of hyperspectral image (HSI) analysis, enabling the extraction of complex and hierarchical features. While convolutional neural networks (CNNs) have been the backbone of HSI classification, their limitations in capturing global contextual features have led to the exploration of Vision Transformers (ViTs). This paper introduces AMBER, an advanced SegFormer specifically designed for multi-band image segmentation. AMBER enhances the original SegFormer by incorporating three-dimensional convolutions to handle hyperspectral data. Our experiments, conducted on the Indian Pines, Pavia University, and PRISMA datasets, show that AMBER outperforms traditional CNN-based methods in terms of Overall Accuracy, Kappa coefficient, and Average Accuracy on the first two datasets, and achieves state-of-the-art performance on the PRISMA dataset.
We describe the construction of end-to-end jet image classifiers based on simulated low-level detector data to discriminate quark- vs. gluon-initiated jets with high-fidelity simulated CMS Open Data. We highlight the importance of precise spatial information and demonstrate competitive performance to existing state-of-the-art jet classifiers. We further generalize the end-to-end approach to event-level classification of quark vs. gluon di-jet QCD events. We compare the fully end-to-end approach to using hand-engineered features and demonstrate that the end-to-end algorithm is robust against the effects of underlying event and pile-up.
Biologists all over the world use camera traps to monitor biodiversity and wildlife population density. The computer vision community has been making strides towards automating the species classification challenge in camera traps, but it has proven difficult to to apply models trained in one region to images collected in different geographic areas. In some cases, accuracy falls off catastrophically in new region, due to both changes in background and the presence of previously-unseen species. We propose a pipeline that takes advantage of a pre-trained general animal detector and a smaller set of labeled images to train a classification model that can efficiently achieve accurate results in a new region.
This paper analyses how well a Fast Fully Convolutional Network (FastFCN) semantically segments satellite images and thus classifies Land Use/Land Cover(LULC) classes. Fast-FCN was used on Gaofen-2 Image Dataset (GID-2) to segment them in five different classes: BuiltUp, Meadow, Farmland, Water and Forest. The results showed better accuracy (0.93), precision (0.99), recall (0.98) and mean Intersection over Union (mIoU)(0.97) than other approaches like using FCN-8 or eCognition, a readily available software. We presented a comparison between the results. We propose FastFCN to be both faster and more accurate automated method than other existing methods for LULC classification.
This is a study on facial information analysis technology for estimating gender and age, and poses are estimated using a transformation relationship matrix between the camera coordinate system and the world coordinate system for estimating the pose of a face image. Gender classification was relatively simple compared to age estimation, and age estimation was made possible using deep learning-based facial recognition technology. A comparative CNN was proposed to calculate the experimental results using the purchased database and the public database, and deep learning-based gender classification and age estimation performed at a significant level and was more robust to environmental changes compared to the existing machine learning techniques.
This study explores the impact of class outliers on the effectiveness of example-based explainability methods for black-box machine learning models. We reformulate existing explainability evaluation metrics, such as correctness and relevance, specifically for example-based methods, and introduce a new metric, distinguishability. Using these metrics, we highlight the shortcomings of current example-based explainability methods, including those who attempt to suppress class outliers. We conduct experiments on two datasets, a text classification dataset and an image classification dataset, and evaluate the performance of four state-of-the-art explainability methods. Our findings underscore the need for robust techniques to tackle the challenges posed by class outliers.
We describe an approach to learning rich representations for images, that enables simple and effective predictors in a range of vision tasks involving spatially structured maps. Our key idea is to map small image elements to feature representations extracted from a sequence of nested regions of increasing spatial extent. These regions are obtained by "zooming out" from the pixel/superpixel all the way to scene-level resolution, and hence we call these zoom-out features. Applied to semantic segmentation and other structured prediction tasks, our approach exploits statistical structure in the image and in the label space without setting up explicit structured prediction mechanisms, and thus avoids complex and expensive inference. Instead image elements are classified by a feedforward multilayer network with skip-layer connections spanning the zoom-out levels. When used in conjunction with modern neural architectures such as ResNet, DenseNet and NASNet (to which it is complementary) our approach achieves competitive accuracy on segmentation benchmarks. In addition, we propose an approach for learning category-level semantic segmentation purely from image-level classification tag. It exploits localization cues that emerge from training a modified zoom-out architecture tailored for classification tasks, to drive a weakly supervised process that automatically labels a sparse, diverse training set of points likely to belong to classes of interest. Finally, we introduce data-driven regularization functions for the supervised training of CNNs. Our innovation takes the form of a regularizer derived by learning an autoencoder over the set of annotations. This approach leverages an improved representation of label space to inform extraction of features from images
Lightweight Image Classification of Raptor Species for Edge Devices: Rare-Species Dataset Expansion via Video Frame Extraction, Knowledge Distillation, and TensorRT Deployment
본 연구에서는 풍력 발전기 충돌 저감을 위한 실시간 엣지 배포(edge deployment) 관점에서 경량 맹금류 종 분류를 탐구한다. 304M 파라미터 규모의 DINOv2-L을 교사 모델로 활용하여 세 가지 경량 학생 모델(MobileNetV4, ViT-Small, EfficientNet-B0)로 지식 증류(knowledge distillation)를 수행했다. 유사 종 간의 혼동을 줄이기 위해 비디오 프레임 추출을 통해 참수리(Steller's Sea Eagle) 이미지를 463장에서 2,050장으로 늘리는 등 데이터셋을 12,519장으로 확장했다. 비디오 및 원본 이미지 수준에서 샘플을 분리하여 해당 단위의 소스 누출(source leakage)을 방지하는 그룹 분할 환경에서, 3개 학생 모델의 앙상블은 약 8분의 1 수준의 파라미터만으로 5개 증류 시드에 걸쳐 0.935 ± 0.004의 매크로 재현율(macro recall)을 달성했다(기존 이미지 수준 분할에서는 0.955를 기록하며 교사 모델 매크로 재현율의 97.5%를 유지함). 이전 학습 이미지와 겹치지 않는 1,258장의 하위 집합에서 흰꼬리수리(White-tailed Eagle)의 재현율은 최대 38.6%p 향상되었으며, 참수리로 오분류되는 비율은 전체 오류의 61%에서 15%로 감소했다. NVIDIA Jetson Orin Nano에서 EfficientNet-B0를 TensorRT FP16으로 배포했을 때 호스트-디바이스 전송을 포함하여 이미지당 3.19ms(초당 313장)의 속도를 기록했으며, FP32와의 argmax 일치도는 99.95%에 달했다. 5개 시드 기반의 통제 비교 실험 결과, 지식 증류(교차 엔트로피 단독 대비)나 DINOv2-L에서 DINOv3-L 교사 모델로의 변경 모두 앙상블 수준에서 뚜렷한 성능 향상을 이끌어내지 못했으며, 주된 성능 개선은 데이터셋 확장과 교사 모델 재미세조정(re-fine-tuning)에서 기인함을 확인했다.
Due to the high volume of traffic on modern roadways, transportation agencies have proposed High Occupancy Vehicle (HOV) lanes and High Occupancy Tolling (HOT) lanes to promote car pooling. However, enforcement of the rules of these lanes is currently performed by roadside enforcement officers using visual observation. Manual roadside enforcement is known to be inefficient, costly, potentially dangerous, and ultimately ineffective. Violation rates up to 50%-80% have been reported, while manual enforcement rates of less than 10% are typical. Therefore, there is a need for automated vehicle occupancy detection to support HOV/HOT lane enforcement. A key component of determining vehicle occupancy is to determine whether or not the vehicle's front passenger seat is occupied. In this paper, we examine two methods of determining vehicle front seat occupancy using a near infrared (NIR) camera system pointed at the vehicle's front windshield. The first method examines a state-of-the-art deformable part model (DPM) based face detection system that is robust to facial pose. The second method examines state-of- the-art local aggregation based image classification using bag-of-visual-words (BOW) and Fisher vectors (FV). A dataset of 3000 images was collected on a public roadway and is used to perform the comparison. From these experiments it is clear that the image classification approach is superior for this problem.
Fine-grained recognition distinguishes among categories with subtle visual differences. In order to differentiate between these challenging visual categories, it is helpful to leverage additional information. Geolocation is a rich source of additional information that can be used to improve fine-grained classification accuracy, but has been understudied. Our contributions to this field are twofold. First, to the best of our knowledge, this is the first paper which systematically examined various ways of incorporating geolocation information into fine-grained image classification through the use of geolocation priors, post-processing or feature modulation. Secondly, to overcome the situation where no fine-grained dataset has complete geolocation information, we release two fine-grained datasets with geolocation by providing complementary information to existing popular datasets - iNaturalist and YFCC100M. By leveraging geolocation information we improve top-1 accuracy in iNaturalist from 70.1% to 79.0% for a strong baseline image-only model. Comparing several models, we found that best performance was achieved by a post-processing model that consumed the output of the image-only baseline alongside geolocation. However, for a resource-constrained model (MobileNetV2), performance was better with a feature modulation model that trains jointly over pixels and geolocation: accuracy increased from 59.6% to 72.2%. Our work makes a strong case for incorporating geolocation information in fine-grained recognition models for both server and on-device.
This paper proposes a multi-layer neural network structure for few-shot image recognition of novel categories. The proposed multi-layer neural network architecture encodes transferable knowledge extracted from a large annotated dataset of base categories. This architecture is then applied to novel categories containing only a few samples. The transfer of knowledge is carried out at the feature-extraction and the classification levels distributed across the two training stages. In the first-training stage, we introduce the relative feature to capture the structure of the data as well as obtain a low-dimensional discriminative space. Secondly, we account for the variable variance of different categories by using a network to predict the variance of each class. Classification is then performed by computing the Mahalanobis distance to the mean-class representation in contrast to previous approaches that used the Euclidean distance. In the second-training stage, a category-agnostic mapping is learned from the mean-sample representation to its corresponding class-prototype representation. This is because the mean-sample representation may not accurately represent the novel category prototype. Finally, we evaluate the proposed network structure on four standard few-shot image recognition datasets, where our proposed few-shot learning system produces competitive performance compared to previous work. We also extensively studied and analyzed the contribution of each component of our proposed framework.
Diffusion-based data augmentation has emerged as a powerful strategy for addressing data-scarce image classification. Recent pipelines typically adapt a pretrained text-to-image diffusion model to a limited number of class examples and subsequently sample synthetic images using diverse textual prompts. However, in extreme data-scarce regimes such as long-tailed single-sample settings, this approach suffers from a dual failure mode: the generated samples often lack discriminative foreground details crucial for classifier training, and the adapted class prior can bias the early denoising process, causing output images to exhibit similar patterns despite varying text prompts. To overcome these limitations, we propose a sampling-side framework of two complementary components that simultaneously enhance fine details and diversity. First, Mask-Guided Reference (MGR) Attention updates only foreground value tokens with reference features via zero-shot segmentation, limiting reference influence to the target object and mitigating excessive global feature injection. Second, Repulsive Noise Optimization (RNO) encourages initial noise trajectories to diverge from dominant generative modes of the adapted model, enabling target prompts to express broader contextual variation. Extensive evaluations show consistent improvements, with the largest gains in long-tailed single-sample settings of up to 2.50 and 2.94 points on CUB-LT IF100, and further robustness across few-shot and background-shift evaluations.
This paper presents an approach for semantic place categorization using data obtained from RGB cameras. Previous studies on visual place recognition and classification have shown that, by considering features derived from pre-trained Convolutional Neural Networks (CNNs) in combination with part-based classification models, high recognition accuracy can be achieved, even in presence of occlusions and severe viewpoint changes. Inspired by these works, we propose to exploit local deep representations, representing images as set of regions applying a Na\"{i}ve Bayes Nearest Neighbor (NBNN) model for image classification. As opposed to previous methods where CNNs are merely used as feature extractors, our approach seamlessly integrates the NBNN model into a fully-convolutional neural network. Experimental results show that the proposed algorithm outperforms previous methods based on pre-trained CNN models and that, when employed in challenging robot place recognition tasks, it is robust to occlusions, environmental and sensor changes.
This manuscript describes our participation in the International Skin Imaging Collaboration's 2017 Skin Lesion Analysis Towards Melanoma Detection competition. We participated in Part 3: Lesion Classification. The two stated goals of this binary image classification challenge were to distinguish between (a) melanoma and (b) nevus and seborrheic keratosis, followed by distinguishing between (a) seborrheic keratosis and (b) nevus and melanoma. We chose a deep neural network approach with a transfer learning strategy, using a pre-trained Inception V3 network as both a feature extractor to provide input for a multi-layer perceptron as well as fine-tuning an augmented Inception network. This approach yielded validation set AUC's of 0.84 on the second task and 0.76 on the first task, for an average AUC of 0.80. We joined the competition unfortunately late, and we look forward to improving on these results.