Image Classification Datasets Research Articles

Semantic Knowledge Guided Class-Incremental Learning

Driven by practical needs, research on Class-Incremental Learning (CIL) has received more and more attentions in recent years. A technical challenge to be conquered by CIL methods is the catastrophic forgetting problem, where the model’s performance improves rapidly on new classes while deteriorates drastically on old ones. The main causes behind catastrophic forgetting include network drifts, inter-class confusions, etc. In this paper, we propose a novel CIL method that solves the catastrophic forgetting problem from two aspects. First, to solve the inter-class confusion problem, we propose a novel Semantic knOwledge gUided ciL framework (SOUL) that consists of a CNN feature extractor and a Bi-GCN (Graph Convolutional Network) classifier. In each CIL session, we use the semantic knowledge extracted from the class labels to build two inter-class relation graphs among all the encountered old and new classes. Using these two relation graphs, we develop a Bi-GCN classifier to fuse two kinds of semantic relations in a balanced way, and then to transfer the inter-class relations from semantic modality to image classification weights. The entire SOUL framework is trained end-to-end by the standard BP algorithm, which optimizes the Bi-GCN classifier and the CNN feature extractor jointly. Second, to prevent the network drift, we develop the local topology preserving strategy that divides the global topological structure of the learned feature space into a set of local topological relations, and maintains these local relations at CIL session. Experimental evaluations demonstrate the state-of-the-art performance accuracies on benchmark image classification datasets.

Nitrogen deficiency in maize: Annotated image classification dataset.

Nitrogen (N) is one of the key inputs in maize production applied in the form of fertilizers. Nitrogen deficiency during the vegetation period leads to lower yields since N is utilized in proteins and enzymes that enable important biochemical processes such as photosynthesis. Nitrogen deficiency leads to specific symptoms that eventually become visible to the naked eye during vegetation. Our hypothesis was that N deficiency can be detected from maize RGB images in parametric process such as a deep neural network. The aim of the reported dataset is to optimize the usage of N in the farmer's fields and accordingly, reduce its environmental footprint. This dataset contains 1200 images of maize canopy from field trials, annotated by an expert from an agricultural institution. The field trials included three levels of N fertilization: N0 without N fertilization, N75 with 75kg of added N fertilizer, and NFull with 136kg of added N fertilizer. For each fertilizer level, 400 plots were created with 238 different maize genotypes, resulting in a total of 1200 plots. Images were taken with a tripod mounted DSLR camera, aperture priority set to f/8 and sensor sensitivity set to ISO400. Images were taken at a 45° angle to each plot. This dataset can be useful to both researchers, data scientists and agronomists, especially in the context of emerging technologies in precision agriculture, such as robotics, 5G networks and unmanned aerial vehicle (UAV). The dataset is one of the first publicly accessible datasets of maize canopy images under different N fertilization levels and represents a valuable public resource for development of machine learning models for in-season detection of N deficiency in maize.

Comparation of Federated and Centralized Learning for Image Classification

Federated Learning (FL) is a new approach in machine learning or it can also be called collaborative learning, which is a machine learning method that includes client devices to carry out the training process, so that clients do not need to send training data to the server but directly conduct training on their respective devices. respectively. The models generated from local training will be sent to the server for further global aggregation. Therefore, FL is referred to as machine learning which can maintain the privacy of the data owner, because the data is not submitted to the server and is still stored in each client's device. In this study, a performance comparison will be carried out to prove whether the latest approach which is Federated Learning, can produce the same accuracy performance as the traditional approach, that is Centralized Machine Learning, in the case of Image Classification. The comparison of two approaches would be conducted by using the Open Source Image Classification dataset, namely MNIST. The performance of two approaches would be presented by evaluation that is Accuracy. The result shows that Federated Learning almost overcome the performance of Centralized Learning in the case of Image Classification by provided Accuracy 76%.

Transformer-BLS: An efficient learning algorithm based on multi-head attention mechanism and incremental learning algorithms

Due to its efficient model calibration given by unique incremental learning capability, broad learning system (BLS) has made impressive progress in image analytical tasks such as image classification and object detection. Inspired by this incremental remodel success, we proposed a novel transformer-BLS network to achieve a trade-off between model training speed and accuracy. Specially, we developed sub-BLS layers with the multi-head attention mechanism and combining these layers to construct a transformer-BLS network. In particular, our proposed transformer-BLS network provides four different incremental learning algorithms that enable the proposed model can realize the increments of its feature nodes, enhancement nodes, input data and sub-BLS layers, respectively, without the need of the full-weight update in this model. Furthermore, we validated the performance of our transformer-BLS network and its four incremental learning algorithms on a variety of image classification datasets. The results demonstrated that the proposed transformer-BLS maintains classification performance on both the MNIST and Fashion-MNIST datasets, while saving 2/3 of the training time. These findings imply that the proposed method has the potential in significant reducing model training complexity with this incremental remodel system, while simultaneously improving the increment learning performance of the original BLS within such contexts, especially in the classification task of some datasets.

Channel-Correlation-Based Selective Knowledge Distillation

As a simple yet effective model compression method, knowledge distillation (or KD) is used to learn a small lightweight student network by transferring valuable knowledge from a pre-trained cumbersome teacher network. However, existing KD methods usually consider the feature knowledge either in different layers or individual samples, failing to explore more detailed information in different channels from the perspective of sample relationships. Meanwhile, the negative influences contained in the teacher knowledge are also not well investigated especially when using the response-based knowledge. To address the above-mentioned issues, we devise a novel knowledge distillation approach entitled channel correlation-based selective knowledge distillation (or CCSKD). Specifically, to distill rich knowledge from feature representations, we not only consider the feature knowledge from different channels for individual samples, but also take into account the relational knowledge based on per-channel features for different samples. Furthermore, to further distill positive response-based knowledge, a selective strategy is developed, i.e., selective knowledge distillation, to progressively correct the negative influences from the teacher knowledge during the distillation process. We perform extensive experiments on three image classification datasets, CIFAR-100, Stanford Cars, and Tiny-ImageNet, to demonstrate the effectiveness of the proposed CCSKD, which outperforms recent state-of-the-art methods with a clear margin. Our codes are publicly available at https://github.com/gjplab/CCSKD.

BookKD: A novel knowledge distillation for reducing distillation costs by decoupling knowledge generation and learning

Knowledge distillation guides student networks’ training and enhances their performance through excellent teacher networks. However, along with the performance advantages, knowledge distillation also entails a huge computational burden, sometimes tens or even hundreds of times that of traditional training methods. So, this paper designs a book-based knowledge distillation (BookKD) to minimize the costs of knowledge distillation while improving performance. First, a decoupling-based knowledge distillation framework is designed. By decoupling the traditional knowledge distillation process into two independent sub-processes, book-making and book-learning, knowledge distillation can be completed with little resource consumption. Second, a book-making method based on knowledge ensemble and knowledge regularization is developed, which makes books by organizing and processing the knowledge generated by teachers. These books can replace these teachers to provide sufficient knowledge with little distillation costs. Finally, a book-learning method based on entropy dynamic adjustment and label smoothing is designed. The entropy dynamic adjustment optimizes the training loss and mitigates student networks’ difficulty in learning books. Label smoothing alleviates the student network’s over-confidence in ground truth labels, which increases its attention to the class similarity knowledge in books. BookKD is tested on three image classification datasets, CIFAR100, ImageNet and ImageNet100, and an object detection dataset PASCAL VOC 2007. The experiment results indicate the advantages of BookKD in reducing distillation costs and improving distillation performance.

Source-Free Progressive Graph Learning for Open-Set Domain Adaptation.

Open-set domain adaptation (OSDA) aims to transfer knowledge from a label-rich source domain to a label-scarce target domain while addressing disturbances from irrelevant target classes not present in the source data. However, most OSDA approaches are limited due to the lack of essential theoretical analysis of generalization bound, reliance on the coexistence of source and target data during adaptation, and failure to accurately estimate model predictions' uncertainty. To address these limitations, the Progressive Graph Learning (PGL) framework is proposed. PGL decomposes the target hypothesis space into shared and unknown subspaces and progressively pseudo-labels the most confident known samples from the target domain for hypothesis adaptation. PGL guarantees a tight upper bound of the target error by integrating a graph neural network with episodic training and leveraging adversarial learning to close the gap between the source and target distributions. The proposed approach also tackles a more realistic source-free open-set domain adaptation (SF-OSDA) setting that makes no assumptions about the coexistence of source and target domains. In a two-stage framework, the SF-PGL model' uniformly selects the most confident target instances from each category at a fixed ratio, and the confidence thresholds in each class weigh the classification loss in the adaptation step. The proposed methods are evaluated on benchmark image classification and action recognition datasets, where they demonstrate superiority and flexibility in recognizing both shared and unknown categories. Additionally, balanced pseudo-labeling plays a significant role in improving calibration, making the trained model less prone to over- or under-confident predictions on the target data.

A surrogate-assisted highly cooperative coevolutionary algorithm for hyperparameter optimization in deep convolutional neural networks

Hyperparameter optimization in convolutional neural networks (CNNs) plays a vital role in ensuring the effectiveness of the models. However, with the depth of the existing CNN expanding, this task becomes very challenging due to the high-dimensional and computationally expensive characteristics. Given these difficulties, this study proposes a surrogate-assisted highly cooperative hyperparameter optimization (SHCHO) algorithm for large-scale chain-styled CNNs. SHCHO tackles the original hyperparameter optimization problem by cooperatively optimizing its subproblems. Specifically, it first decomposes the whole CNN into several overlapping sub-CNNs, which conforms to the inherent overlapping interaction structure among hyperparameters and significantly reduces the search space. The resulting hyperparameter optimization subproblems on these sub-CNNs are then coevolved collaboratively and competitively, facilitating a proper hyperparameter configuration for the whole CNN. Besides, SHCHO also employs the computationally efficient surrogate technique to assist in each subproblem optimization. As a result, the expensive computational cost can be greatly reduced. Extensive experimental results on three widely-used image classification datasets indicate that SHCHO shows competitive performance when compared with several state-of-the-art algorithms in terms of both hyperparameter superiority and computational cost.

Single-shot pruning and quantization for hardware-friendly neural network acceleration

Applying CNN on embedded systems is challenging due to model size limitations. Pruning and quantization can help, but are time-consuming to apply separately. Our Single-Shot Pruning and Quantization strategy addresses these issues by quantizing and pruning in a single process. We evaluated our method on CIFAR-10 and CIFAR-100 datasets for image classification. Our model is 69.4% smaller with little accuracy loss, and runs 6–8 times faster on NVIDIA Xavier NX hardware.

DiRaC-I: Identifying Diverse and Rare Training Classes for Zero-Shot Learning

Zero-Shot Learning (ZSL) is an extreme form of transfer learning that aims at learning from a few “seen classes” to have an understanding about the “unseen classes” in the wild. Given a dataset in ZSL research, most existing works use a predetermined, disjoint set of seen-unseen classes to evaluate their methods. These seen (training) classes might be sub-optimal for ZSL methods to appreciate the diversity and rarity of an object domain. Inspired by strategies like active learning, it is intuitive that intelligently selecting the training classes can improve ZSL performance. In this work, we propose a framework called Diverse and Rare Class Identifier (DiRaC-I) which, given an attribute-based dataset, can intelligently yield the most suitable “seen classes” for training ZSL models. DiRaC-I has two main goals – constructing a diversified set of seed classes, and using them to initialize a visual-semantic mining algorithm for acquiring the classes capturing both diversity and rarity in the object domain adequately. These classes can then be used as “seen classes” to train ZSL models for image classification. We simulate a real-world scenario where visual samples of novel object classes in the wild are available to neither DiRaC-I nor the ZSL models during training and conducted extensive experiments on two benchmark data sets for zero-shot image classification — CUB and SUN. Our results demonstrate DiRaC-I helps ZSL models to achieve significant classification accuracy improvements – specifically, up to 8% for CUB and up to 5% for SUN dataset. Additionally, while recognizing classes exhibiting rare attributes we also observe a performance boost for ZSL models, which is up to 10% and 7% for CUB and SUN datasets, respectively.

Scale-preserving automatic concept extraction (SPACE)

Convolutional Neural Networks (CNN) have become a common choice for industrial quality control, as well as other critical applications in the Industry 4.0. When these CNNs behave in ways unexpected to human users or developers, severe consequences can arise, such as economic losses or an increased risk to human life. Concept extraction techniques can be applied to increase the reliability and transparency of CNNs through generating global explanations for trained neural network models. The decisive features of image datasets in quality control often depend on the feature’s scale; for example, the size of a hole or an edge. However, existing concept extraction methods do not correctly represent scale, which leads to problems interpreting these models as we show herein. To address this issue, we introduce the Scale-Preserving Automatic Concept Extraction (SPACE) algorithm, as a state-of-the-art alternative concept extraction technique for CNNs, focused on industrial applications. SPACE is specifically designed to overcome the aforementioned problems by avoiding scale changes throughout the concept extraction process. SPACE proposes an approach based on square slices of input images, which are selected and then tiled before being clustered into concepts. Our method provides explanations of the models’ decision-making process in the form of human-understandable concepts. We evaluate SPACE on three image classification datasets in the context of industrial quality control. Through experimental results, we illustrate how SPACE outperforms other methods and provides actionable insights on the decision mechanisms of CNNs. Finally, code for the implementation of SPACE is provided.

Guiding the retraining of convolutional neural networks against adversarial inputs.

When using deep learning models, one of the most critical vulnerabilities is their exposure to adversarial inputs, which can cause wrong decisions (e.g., incorrect classification of an image) with minor perturbations. To address this vulnerability, it becomes necessary to retrain the affected model against adversarial inputs as part of the software testing process. In order to make this process energy efficient, data scientists need support on which are the best guidance metrics for reducing the adversarial inputs to create and use during testing, as well as optimal dataset configurations. We examined six guidance metrics for retraining deep learning models, specifically with convolutional neural network architecture, and three retraining configurations. Our goal is to improve the convolutional neural networks against the attack of adversarial inputs with regard to the accuracy, resource utilization and execution time from the point of view of a data scientist in the context of image classification. We conducted an empirical study using five datasets for image classification. We explore: (a) the accuracy, resource utilization, and execution time of retraining convolutional neural networks with the guidance of six different guidance metrics (neuron coverage, likelihood-based surprise adequacy, distance-based surprise adequacy, DeepGini, softmax entropy and random), (b) the accuracy and resource utilization of retraining convolutional neural networks with three different configurations (one-step adversarial retraining, adversarial retraining and adversarial fine-tuning). We reveal that adversarial retraining from original model weights, and by ordering with uncertainty metrics, gives the best model w.r.t. accuracy, resource utilization, and execution time. Although more studies are necessary, we recommend data scientists use the above configuration and metrics to deal with the vulnerability to adversarial inputs of deep learning models, as they can improve their models against adversarial inputs without using many inputs and without creating numerous adversarial inputs. We also show that dataset size has an important impact on the results.

Studies Advanced in Image Classification based on Deep Learning

Accurate and efficient image classification is one of the important research topics in image analysis, and it has also been a research hotspot in the computer vision community.Deep neural networks are increasingly being used for picture categorization and processing in recent years as a result of advancements in machine learning technology.In this article, we introduce the research progress of image recognition technology based on depth learning, including the design ideas, principles, structures, advantages and disadvantages of several depth neural networks. In addition, we quantitatively compared the recognition results of different methods on classic image classification datasets. Lastly, we review some current issues with picture classification and talk about the direction of deep neural network development in the use of image classification technology.

Hierarchical Bias Mitigation for Semi-Supervised Medical Image Classification

Semi-supervised learning (SSL) has demonstrated remarkable advances on medical image classification, by harvesting beneficial knowledge from abundant unlabeled samples. The pseudo labeling dominates current SSL approaches, however, it suffers from intrinsic biases within the process. In this paper, we retrospect the pseudo labeling and identify three hierarchical biases: perception bias, selection bias and confirmation bias, at feature extraction, pseudo label selection and momentum optimization stages, respectively. In this regard, we propose a HierArchical BIas miTigation (HABIT) framework to amend these biases, which consists of three customized modules including Mutual Reconciliation Network (MRNet), Recalibrated Feature Compensation (RFC) and Consistency-aware Momentum Heredity (CMH). Firstly, in the feature extraction, MRNet is devised to jointly utilize convolution and permutator-based paths with a mutual information transfer module to exchanges features and reconcile spatial perception bias for better representations. To address pseudo label selection bias, RFC adaptively recalibrates the strong and weak augmented distributions to be a rational discrepancy and augments features for minority categories to achieve the balanced training. Finally, in the momentum optimization stage, in order to reduce the confirmation bias, CMH models the consistency among different sample augmentations into network updating process to improve the dependability of the model. Extensive experiments on three semi-supervised medical image classification datasets demonstrate that HABIT mitigates three biases and achieves state-of-the-art performance. Our codes are available at https://github.com/ CityU-AIM-Group/HABIT.

Attention Spiking Neural Networks.

Brain-inspired spiking neural networks (SNNs) are becoming a promising energy-efficient alternative to traditional artificial neural networks (ANNs). However, the performance gap between SNNs and ANNs has been a significant hindrance to deploying SNNs ubiquitously. To leverage the full potential of SNNs, in this paper we study the attention mechanisms, which can help human focus on important information. We present our idea of attention in SNNs with a multi-dimensional attention module, which infers attention weights along the temporal, channel, as well as spatial dimension separately or simultaneously. Based on the existing neuroscience theories, we exploit the attention weights to optimize membrane potentials, which in turn regulate the spiking response. Extensive experimental results on event-based action recognition and image classification datasets demonstrate that attention facilitates vanilla SNNs to achieve sparser spiking firing, better performance, and energy efficiency concurrently. In particular, we achieve top-1 accuracy of 75.92% and 77.08% on ImageNet-1 K with single/4-step Res-SNN-104, which are state-of-the-art results in SNNs. Compared with counterpart Res-ANN-104, the performance gap becomes -0.95/+0.21 percent and the energy efficiency is 31.8×/7.4×. To analyze the effectiveness of attention SNNs, we theoretically prove that the spiking degradation or the gradient vanishing, which usually holds in general SNNs, can be resolved by introducing the block dynamical isometry theory. We also analyze the efficiency of attention SNNs based on our proposed spiking response visualization method. Our work lights up SNN's potential as a general backbone to support various applications in the field of SNN research, with a great balance between effectiveness and energy efficiency.

A novel Out-of-Distribution detection approach for Spiking Neural Networks: Design, fusion, performance evaluation and explainability

Research around Spiking Neural Networks has ignited during the last years due to their advantages when compared to traditional neural networks, including their efficient processing and inherent ability to model complex temporal dynamics. Despite these differences, Spiking Neural Networks face similar issues than other neural computation counterparts when deployed in real-world settings. This work addresses one of the practical circumstances that can hinder the trustworthiness of this family of models: the possibility of querying a trained model with samples far from the distribution of its training data (also referred to as Out-of-Distribution or OoD data). Specifically, this work presents a novel OoD detector that can identify whether test examples input to a Spiking Neural Network belong to the distribution of the data over which it was trained. For this purpose, we characterize the internal activations of the hidden layers of the network in the form of spike count patterns, which lay a basis for determining when the activations induced by a test instance is atypical. Furthermore, a local explanation method is devised to produce attribution maps revealing which parts of the input instance push most towards the detection of an example as an OoD sample. Experimental results are performed over several classic and event-based image classification datasets to compare the performance of the proposed detector to that of other OoD detection schemes from the literature. Our experiments also assess whether the fusion of our proposed approach with other baseline OoD detection schemes can complement and boost the overall OoD detection capability As the obtained results clearly show, the proposed detector performs competitively against such alternative schemes, and when fused together, can significantly improve the detection scores of their constituent individual detectors. Furthermore, the explainability technique associated to our proposal is proven to produce relevance attribution maps that conform to expectations for synthetically created OoD instances.

BreastDM: A DCE-MRI dataset for breast tumor image segmentation and classification

Dynamic contrast-enhanced magnetic resonance imaging (DCE-MRI) has shown high sensitivity to diagnose breast cancer. However, few computer-aided algorithms focus on employing DCE-MR images for breast cancer diagnosis due to the lack of publicly available DCE-MRI datasets. To address this issue, our work releases a new DCE-MRI dataset called BreastDM for breast tumor segmentation and classification. In particular, a dataset of 232 patients selected with DCE-MR images for benign and malignant cases is established. Each case consists of three types of sequences: pre-contrast, post-contrast, and subtraction sequences. To show the difficulty of breast DCE-MRI tumor image segmentation and classification tasks, benchmarks are achieved by state-of-the-art image segmentation and classification algorithms, including conventional hand-crafted based methods and recently-emerged deep learning-based methods. More importantly, a local–global cross attention fusion network (LG-CAFN) is proposed to further improve the performance of breast tumor images classification. Specifically, LG-CAFN achieved the highest accuracy (88.20%, 83.93%) and AUC value (0.9154,0.8826) in both groups of experiments. Extensive experiments are conducted to present strong baselines based on various typical image segmentation and classification algorithms. Experiment results also demonstrate the superiority of the proposed LG-CAFN to other breast tumor images classification methods. The related dataset and evaluation codes are publicly available at smallboy-code/Breast-cancer-dataset.

Autonomous perception and adaptive standardization for few-shot learning

Identifying unseen classes with limited labeled data for reference is a challenging task, which is also known as few-shot learning. Generally, a knowledge-rich model is more robust than a knowledge-poor model when facing novel situations, and an intuitive way to enrich knowledge is to find additional training data, but this is not compatible with the principle of few-shot learning which aims to reduce reliance on big data. In contrast, improving the utilization of existing data is a more attractive option. In this paper, we propose a batch perception distillation approach, which improves the utilization of existing data by guiding individual classification with the intermixed information across a batch. In addition to data utilization, obtaining robust feature representation is also a concern. Specifically, the widely adopted metric-based few-shot classification approach classifies unseen testing classes by comparing the extracted features of different novel samples, which requires that the extracted features can accurately represent the class-related clues of the input images. In this paper, we propose a salience perception attention that enables the model to focus more easily on key clues in images, which helps to reduce the interference of irrelevant factors during classification. To overcome the distribution gap between the training classes and the unseen testing classes, we propose a weighted centering post-processing that standardizes the testing data according to the similarity between the training and testing classes. By combining the three proposed components, our method achieves superior performance on four widely used few-shot image classification datasets.

Context-aware deep learning with dynamically assembled weight matrices

Deep neural networks are static by nature, meaning they use a single set of parameters to process each data sample. However, for more complex and larger systems, for which samples can be obtained under a large variety of circumstances, a need for more dynamic networks that adapt to these variations seems apparent. To this end, the concept of context information and its integration in deep learning is considered. Contrary to current practices, that treat all modalities as equally informative to the decision process, contextual and feature information is considered to be vastly different in nature. A set of definitions and subsequent arguments are given in order to provide the necessary clarification regarding the interpretation of context with the purpose of using all information in a maximally efficient way. From this interpretation, a problem statement of context aware deep learning is constructed and consequently linked to its multiple model and transfer learning solutions. These solutions, however, are in-efficient since samples are spread across models. Based on the existence of this multiple model solution, a new approach, which integrates contextual information directly into a single context-dependent model, is proposed. This single model uses weight matrices that are dynamically assembled based on the contextual information to process each data sample individually. This allows for the single model to consume and learn from all samples. The corresponding training routine is constructed and evaluated on multiple benchmark problems. We start with an artificially generated problem on which the methods’ ability to model multiple linear classification problems concurrently is confirmed. Next, both a time series forecasting and image classification dataset are used for evaluation. Evaluations of our proposed method are done and compared to standard context aware implementations based on concatenation and gating. These standard methods implement context information by adding additional parameters in order to try modeling all interactions between the context information and the samples. However, our proposed approach integrates the contextual information directly into the network weights, allowing parameter efficient modeling of dynamic contextual behavior. In both cases the proposed solution outperforms its standard counterparts with significant margins in both evaluation metrics and parameter efficiency. Specifically, a mean absolute error improvement of eleven standard deviations and an eight percent increase in classification accuracy, for the forecasting and image classification problems respectively, is observed, showing the potential of our approach.

IDGAN: Information-Driven Generative Adversarial Network of Coverless Image Steganography

Traditional image steganography techniques complete the steganography process by embedding secret information into cover images, but steganalysis tools can easily detect detectable pixel changes that lead to the leakage of confidential information. The use of a generative adversarial network (GAN) makes it possible to embed information using a combination of information and noise in generating images to achieve steganography. However, this approach is usually accompanied by issues such as poor image quality and low steganography capacity. To address these challenges, we propose a steganography model based on a novel information-driven generative adversarial network (IDGAN), which fuses a GAN, attention mechanisms, and image interpolation techniques. We introduced an attention mechanism on top of the original GAN model to improve image accuracy. In the generation model, we replaced some transposed convolution operations with image interpolation for better quality of dense images. In contrast to traditional steganographic methods, the IDGAN generates images containing confidential information without using cover images and utilizes GANs for information embedding, thus having better anti-detection capability. Moreover, the IDGAN uses an attention mechanism to improve the image details and clarity and optimizes the steganography effect through an image interpolation algorithm. Experimental results demonstrate that the IDGAN achieves an accuracy of 99.4%, 95.4%, 93.2%, and 100% on the MNIST, Intel Image Classification, Flowers, and Face datasets, respectively, with an embedding rate of 0.17 bpp. The model effectively protects confidential information while maintaining high image quality.