Given a training dataset composed of images and corresponding category labels, deep convolutional neural networks show a strong ability in mining discriminative parts for image classification. However, deep convolutional neural networks trained with image level labels only tend to focus on the most discriminative parts while missing other object parts, which could provide complementary information. In this paper, we approach this problem from a different perspective. We build complementary parts models in a weakly supervised manner to retrieve information suppressed by dominant object parts detected by convolutional neural networks. Given image level labels only, we first extract rough object instances by performing weakly supervised object detection and instance segmentation using Mask R-CNN and CRF-based segmentation. Then we estimate and search for the best parts model for each object instance under the principle of preserving as much diversity as possible. In the last stage, we build a bi-directional long short-term memory (LSTM) network to fuze and encode the partial information of these complementary parts into a comprehensive feature for image classification. Experimental results indicate that the proposed method not only achieves significant improvement over our baseline models, but also outperforms state-of-the-art algorithms by a large margin (6.7%, 2.8%, 5.2% respectively) on Stanford Dogs 120, Caltech-UCSD Birds 2011-200 and Caltech 256.
Trust and interpretability are crucial for the use of Artificial Intelligence (AI) in scientific research, but current models often operate as black boxes offering limited transparency and justifications for their outputs. We introduce AiSciVision, a framework that specializes Large Multimodal Models (LMMs) into interactive research partners and classification models for image classification tasks in niche scientific domains. Our framework uses two key components: (1) Visual Retrieval-Augmented Generation (VisRAG) and (2) domain-specific tools utilized in an agentic workflow. To classify a target image, AiSciVision first retrieves the most similar positive and negative labeled images as context for the LMM. Then the LMM agent actively selects and applies tools to manipulate and inspect the target image over multiple rounds, refining its analysis before making a final prediction. These VisRAG and tooling components are designed to mirror the processes of domain experts, as humans often compare new data to similar examples and use specialized tools to manipulate and inspect images before arriving at a conclusion. Each inference produces both a prediction and a natural language transcript detailing the reasoning and tool usage that led to the prediction. We evaluate AiSciVision on three real-world scientific image classification datasets: detecting the presence of aquaculture ponds, diseased eelgrass, and solar panels. Across these datasets, our method outperforms fully supervised models in low and full-labeled data settings. AiSciVision is actively deployed in real-world use, specifically for aquaculture research, through a dedicated web application that displays and allows the expert users to converse with the transcripts. This work represents a crucial step toward AI systems that are both interpretable and effective, advancing their use in scientific research and scientific discovery.
Deep learning models have been successfully deployed for a diverse array of image-based plant phenotyping applications including disease detection and classification. However, successful deployment of supervised deep learning models requires large amount of labeled data, which is a significant challenge in plant science (and most biological) domains due to the inherent complexity. Specifically, data annotation is costly, laborious, time consuming and needs domain expertise for phenotyping tasks, especially for diseases. To overcome this challenge, active learning algorithms have been proposed that reduce the amount of labeling needed by deep learning models to achieve good predictive performance. Active learning methods adaptively select samples to annotate using an acquisition function to achieve maximum (classification) performance under a fixed labeling budget. We report the performance of four different active learning methods, (1) Deep Bayesian Active Learning (DBAL), (2) Entropy, (3) Least Confidence, and (4) Coreset, with conventional random sampling-based annotation for two different image-based classification datasets. The first image dataset consists of soybean [Glycine max L. (Merr.)] leaves belonging to eight different soybean stresses and a healthy class, and the second consists of nine different weed species from the field. For a fixed labeling budget, we observed that the classification performance of deep learning models with active learning-based acquisition strategies is better than random sampling-based acquisition for both datasets. The integration of active learning strategies for data annotation can help mitigate labelling challenges in the plant sciences applications particularly where deep domain knowledge is required.
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.
Nonlinear methods such as Deep Neural Networks (DNNs) are the gold standard for various challenging machine learning problems, e.g., image classification, natural language processing or human action recognition. Although these methods perform impressively well, they have a significant disadvantage, the lack of transparency, limiting the interpretability of the solution and thus the scope of application in practice. Especially DNNs act as black boxes due to their multilayer nonlinear structure. In this paper we introduce a novel methodology for interpreting generic multilayer neural networks by decomposing the network classification decision into contributions of its input elements. Although our focus is on image classification, the method is applicable to a broad set of input data, learning tasks and network architectures. Our method is based on deep Taylor decomposition and efficiently utilizes the structure of the network by backpropagating the explanations from the output to the input layer. We evaluate the proposed method empirically on the MNIST and ILSVRC data sets.
Convolutional neural networks (CNNs) attained a good performance in hyperspectral sensing image (HSI) classification, but CNNs consider spectra as orderless vectors. Therefore, considering the spectra as sequences, recurrent neural networks (RNNs) have been applied in HSI classification, for RNNs is skilled at dealing with sequential data. However, for a long-sequence task, RNNs is difficult for training and not as effective as we expected. Besides, spatial contextual features are not considered in RNNs. In this study, we propose a Shorten Spatial-spectral RNN with Parallel-GRU (St-SS-pGRU) for HSI classification. A shorten RNN is more efficient and easier for training than band-by-band RNN. By combining converlusion layer, the St-SSpGRU model considers not only spectral but also spatial feature, which results in a better performance. An architecture named parallel-GRU is also proposed and applied in St-SS-pGRU. With this architecture, the model gets a better performance and is more robust.
3D Swin Transformer (3D-ST) known for its hierarchical attention and window-based processing, excels in capturing intricate spatial relationships within images. Spatial-spectral Transformer (SST), meanwhile, specializes in modeling long-range dependencies through self-attention mechanisms. Therefore, this paper introduces a novel method: an attentional fusion of these two transformers to significantly enhance the classification performance of Hyperspectral Images (HSIs). What sets this approach apart is its emphasis on the integration of attentional mechanisms from both architectures. This integration not only refines the modeling of spatial and spectral information but also contributes to achieving more precise and accurate classification results. The experimentation and evaluation of benchmark HSI datasets underscore the importance of employing disjoint training, validation, and test samples. The results demonstrate the effectiveness of the fusion approach, showcasing its superiority over traditional methods and individual transformers. Incorporating disjoint samples enhances the robustness and reliability of the proposed methodology, emphasizing its potential for advancing hyperspectral image classification.
Unmanned Aerial Vehicles (UAVs), equipped with camera sensors can facilitate enhanced situational awareness for many emergency response and disaster management applications since they are capable of operating in remote and difficult to access areas. In addition, by utilizing an embedded platform and deep learning UAVs can autonomously monitor a disaster stricken area, analyze the image in real-time and alert in the presence of various calamities such as collapsed buildings, flood, or fire in order to faster mitigate their effects on the environment and on human population. To this end, this paper focuses on the automated aerial scene classification of disaster events from on-board a UAV. Specifically, a dedicated Aerial Image Database for Emergency Response (AIDER) applications is introduced and a comparative analysis of existing approaches is performed. Through this analysis a lightweight convolutional neural network (CNN) architecture is developed, capable of running efficiently on an embedded platform achieving ~3x higher performance compared to existing models with minimal memory requirements with less than 2% accuracy drop compared to the state-of-the-art. These preliminary results provide a solid basis for further experimentation towards real-time aerial image classification for emergency response applications using UAVs.
Modern deep learning approaches usually utilize modality-specific processing. For example, the most common deep learning approach to image classification involves decoding image file bytes into an RGB tensor which is passed into a neural network. Instead, we investigate modality-independent representation learning by performing classification directly on file bytes, without the need for decoding files at inference time. This enables models to operate on various modalities without any hand-designed, modality-specific processing. Our model, ByteFormer, improves ImageNet Top-1 classification accuracy by $5\%$ (from $72.2\%$ to $77.33\%$) relative to DeIT models of similar size. Compared to Perceiver IO, our model requires absolutely no modality-specific processing at inference time, and uses an order of magnitude fewer parameters at equivalent accuracy on ImageNet. We demonstrate that the same ByteFormer architecture can perform audio classification without modifications or modality-specific preprocessing. We achieve $95.42\%$ classification accuracy on the Speech Commands V2 dataset (comparable to the state-of-the-art accuracy of $98.7\%$). Additionally, we demonstrate that ByteFormer can operate jointly on images and audio, handling joint classification without explicit knowledge of the input modality. We release our code at https://github.com/apple/corenet/tree/main/projects/byteformer.
Mammogram benign or malignant classification with only image-level labels is challenging due to the absence of lesion annotations. Motivated by the symmetric prior that the lesions on one side of breasts rarely appear in the corresponding areas on the other side, given a diseased image, we can explore a counterfactual problem that how would the features have behaved if there were no lesions in the image, so as to identify the lesion areas. We derive a new theoretical result for counterfactual generation based on the symmetric prior. By building a causal model that entails such a prior for bilateral images, we obtain two optimization goals for counterfactual generation, which can be accomplished via our newly proposed counterfactual generative network. Our proposed model is mainly composed of Generator Adversarial Network and a \emph{prediction feedback mechanism}, they are optimized jointly and prompt each other. Specifically, the former can further improve the classification performance by generating counterfactual features to calculate lesion areas. On the other hand, the latter helps counterfactual generation by the supervision of classification loss. The utility of our method and the effectiveness of each module in our model can be verified by state-of-the-art performance on INBreast and an in-house dataset and ablation studies.
Deep learning methods have been successfully applied to hyperspectral image (HSI) classification with remarkable performance. Because of limited labelled HSI data, earlier studies primarily adopted a patch-based classification framework, which divides images into overlapping patches for training and testing. However, this approach results in redundant computations and possible information leakage. In this study, we propose an objective evaluation-based high-efficiency learning framework for tiny HSI classification. This framework comprises two main parts: (i) a leakage-free balanced sampling strategy, and (ii) a modified end-to-end fully convolutional network (FCN) architecture that optimizes the trade-off between accuracy and efficiency. The leakage-free balanced sampling strategy generates balanced and non-overlapping training and testing data by partitioning an HSI and the ground truth image into small windows, each of which corresponds to one training or testing sample. The proposed high-efficiency FCN exhibits a pixel-to-pixel architecture with modifications aimed at faster inference speed and improved parameter efficiency. Experiments conducted on four representative datasets demonstrated that the proposed sampling strategy can provide objective performance evaluation and that the proposed network outperformed many state-of-the-art approaches with respect to the speed/accuracy tradeoff. Our source code is available at https://github.com/xmzhang2018.
Contrastive learning (CL) has shown impressive advances in image representation learning in whichever supervised multi-class classification or unsupervised learning. However, these CL methods fail to be directly adapted to multi-label image classification due to the difficulty in defining the positive and negative instances to contrast a given anchor image in multi-label scenario, let the label missing one alone, implying that borrowing a commonly-used way from contrastive multi-class learning to define them will incur a lot of false negative instances unfavorable for learning. In this paper, with the introduction of a label correction mechanism to identify missing labels, we first elegantly generate positives and negatives for individual semantic labels of an anchor image, then define a unique contrastive loss for multi-label image classification with missing labels (CLML), the loss is able to accurately bring images close to their true positive images and false negative images, far away from their true negative images. Different from existing multi-label CL losses, CLML also preserves low-rank global and local label dependencies in the latent representation space where such dependencies have been shown to be helpful in dealing with missing labels. To the best of our knowledge, this is the first general multi-label CL loss in the missing-label scenario and thus can seamlessly be paired with those losses of any existing multi-label learning methods just via a single hyperparameter. The proposed strategy has been shown to improve the classification performance of the Resnet101 model by margins of 1.2%, 1.6%, and 1.3% respectively on three standard datasets, MSCOCO, VOC, and NUS-WIDE. Code is available at https://github.com/chuangua/ContrastiveLossMLML.
We evaluated whether model explanations could efficiently detect bias in image classification by highlighting discriminating features, thereby removing the reliance on sensitive attributes for fairness calculations. To this end, we formulated important characteristics for bias detection and observed how explanations change as the degree of bias in models change. The paper identifies strengths and best practices for detecting bias using explanations, as well as three main weaknesses: explanations poorly estimate the degree of bias, could potentially introduce additional bias into the analysis, and are sometimes inefficient in terms of human effort involved.
We propose a novel framework named ViOCE that integrates ontology-based background knowledge in the form of $n$-ball concept embeddings into a neural network based vision architecture. The approach consists of two components - converting symbolic knowledge of an ontology into continuous space by learning n-ball embeddings that capture properties of subsumption and disjointness, and guiding the training and inference of a vision model using the learnt embeddings. We evaluate ViOCE using the task of few-shot image classification, where it demonstrates superior performance on two standard benchmarks.
We present ResMLP, an architecture built entirely upon multi-layer perceptrons for image classification. It is a simple residual network that alternates (i) a linear layer in which image patches interact, independently and identically across channels, and (ii) a two-layer feed-forward network in which channels interact independently per patch. When trained with a modern training strategy using heavy data-augmentation and optionally distillation, it attains surprisingly good accuracy/complexity trade-offs on ImageNet. We also train ResMLP models in a self-supervised setup, to further remove priors from employing a labelled dataset. Finally, by adapting our model to machine translation we achieve surprisingly good results. We share pre-trained models and our code based on the Timm library.
We propose a versatile deep image compression network based on Spatial Feature Transform (SFT arXiv:1804.02815), which takes a source image and a corresponding quality map as inputs and produce a compressed image with variable rates. Our model covers a wide range of compression rates using a single model, which is controlled by arbitrary pixel-wise quality maps. In addition, the proposed framework allows us to perform task-aware image compressions for various tasks, e.g., classification, by efficiently estimating optimized quality maps specific to target tasks for our encoding network. This is even possible with a pretrained network without learning separate models for individual tasks. Our algorithm achieves outstanding rate-distortion trade-off compared to the approaches based on multiple models that are optimized separately for several different target rates. At the same level of compression, the proposed approach successfully improves performance on image classification and text region quality preservation via task-aware quality map estimation without additional model training. The code is available at the project website: https://github.com/micmic123/QmapCompression
Learning from a limited number of labeled samples (pixels) remains a key challenge in the hyperspectral image (HSI) classification. To address this issue, we propose a deep metric learning-based feature embedding model, which can meet the tasks both for same- and cross-scene HSI classifications. In the first task, when only a few labeled samples are available, we employ ideas from metric learning based on deep embedding features and make a similarity learning between pairs of samples. In this case, the proposed model can learn well to compare whether two samples belong to the same class. In another task, when an HSI image (target scene) that needs to be classified is not labeled at all, the embedding model can learn from another similar HSI image (source scene) with sufficient labeled samples and then transfer to the target model by using an unsupervised domain adaptation technique, which not only employs the adversarial approach to make the embedding features from the source and target samples indistinguishable but also encourages the target scene's embeddings to form similar clusters with the source scene one. After the domain adaptation between the HSIs of the two scenes is finished, any traditional HSI classifier can be used. In a simple manner, the nearest neighbor (NN) algorithm is selected as the classifier for the classification tasks throughout this article. The experimental results from a series of popular HSIs demonstrate the advantages of the proposed model both in the same- and cross-scene classification tasks.
Recently, there has been a growing interest in developing Computer Aided Diagnostic (CAD) systems for improving the reliability and consistency of pathology test results. This paper describes a novel CAD system for the Anti-Nuclear Antibody (ANA) test via Indirect Immunofluorescence protocol on Human Epithelial Type 2 (HEp-2) cells. While prior works have primarily focused on classifying cell images extracted from ANA specimen images, this work takes a further step by focussing on the specimen image classification problem itself. Our system is able to efficiently classify specimen images as well as producing meaningful descriptions of ANA pattern class which helps physicians to understand the differences between various ANA patterns. We achieve this goal by designing a specimen-level image descriptor that: (1) is highly discriminative; (2) has small descriptor length and (3) is semantically meaningful at the cell level. In our work, a specimen image descriptor is represented by its overall cell attribute descriptors. As such, we propose two max-margin based learning schemes to discover cell attributes whilst still maintaining the discrimination of the specimen image descriptor. Our learning schemes differ from the existing discriminative attribute learning approaches as they primarily focus on discovering image-level attributes. Comparative evaluations were undertaken to contrast the proposed approach to various state-of-the-art approaches on a novel HEp-2 cell dataset which was specifically proposed for the specimen-level classification. Finally, we showcase the ability of the proposed approach to provide textual descriptions to explain ANA patterns.
We propose a highly data-efficient active learning framework for image classification. Our novel framework combines: (1) unsupervised representation learning of a Convolutional Neural Network and (2) the Gaussian Process (GP) method, in sequence to achieve highly data and label efficient classifications. Moreover, both elements are less sensitive to the prevalent and challenging class imbalance issue, thanks to the (1) feature learned without labels and (2) the Bayesian nature of GP. The GP-provided uncertainty estimates enable active learning by ranking samples based on the uncertainty and selectively labeling samples showing higher uncertainty. We apply this novel combination to the severely imbalanced case of COVID-19 chest X-ray classification and the Nerthus colonoscopy classification. We demonstrate that only . 10% of the labeled data is needed to reach the accuracy from training all available labels. We also applied our model architecture and proposed framework to a broader class of datasets with expected success.
Concept Bottleneck Models (CBMs) decompose image classification into a process governed by interpretable, human-readable concepts. Recent advances in CBMs have used Large Language Models (LLMs) to generate candidate concepts. However, a critical question remains: What is the optimal number of concepts to use? Current concept banks suffer from redundancy or insufficient coverage. To address this issue, we introduce a dynamic, agent-based approach that adjusts the concept bank in response to environmental feedback, optimizing the number of concepts for sufficiency yet concise coverage. Moreover, we propose Conditional Concept Bottleneck Models (CoCoBMs) to overcome the limitations in traditional CBMs' concept scoring mechanisms. It enhances the accuracy of assessing each concept's contribution to classification tasks and feature an editable matrix that allows LLMs to correct concept scores that conflict with their internal knowledge. Our evaluations across 6 datasets show that our method not only improves classification accuracy by 6% but also enhances interpretability assessments by 30%.