Discriminative Fisher Embedding Dictionary Transfer Learning for Object Recognition

Fan, Zizhu; Shi, Linrui; Liu, Qiang; Li, Zhengming; Zhang, Zheng

doi:10.1109/tnnls.2021.3089566

Cited by 13 publications

(6 citation statements)

References 50 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Automating image annotation, or at least offering automated assistance in this process, can speed up the work and increase its accuracy and consistency [20]. The application of techniques designed to maximize learning from limited data such as transfer learning [21,22,23,24,25], data augmentation [26,27,28,29] and few-shot learning [30,31,32,33,34] complements this move towards automation. While these methods are valuable for training robust models with sparse annotated datasets, the ultimate goal remains to minimize their need by improving the automation of the annotation process itself.…”

Section: A Motivationmentioning

confidence: 99%

Enhancing Image Annotation With Object Tracking and Image Retrieval: A Systematic Review

Fernandes,

Pessoa,

Salgado

et al. 2024

IEEE Access

View full text Add to dashboard Cite

Effective image and video annotation is a fundamental pillar in computer vision and artificial intelligence, crucial for the development of accurate machine learning models. Object tracking and image retrieval techniques are essential in this process, significantly improving the efficiency and accuracy of automatic annotation. This paper systematically investigates object tracking and image acquisition techniques. It explores how these technologies can collectively enhance the efficiency and accuracy of the annotation processes for image and video datasets. Object tracking is examined for its role in automating annotations by tracking objects across video sequences, while image retrieval is evaluated for its ability to suggest annotations for new images based on existing data. The review encompasses diverse methodologies, including advanced neural networks and machine learning techniques, highlighting their effectiveness in various contexts like medical analyses and urban monitoring. Despite notable advancements, challenges such as algorithm robustness and effective human-AI collaboration are identified. This review provides valuable insights into these technologies' current state and future potential in improving image annotation processes, even showing existing applications of these techniques and their full potential when combined.

show abstract

Section: A Motivationmentioning

confidence: 99%

Enhancing Image Annotation With Object Tracking and Image Retrieval: A Systematic Review

Fernandes,

Pessoa,

Salgado

et al. 2024

IEEE Access

View full text Add to dashboard Cite

show abstract

“…Supervised hashing-based multi-modal methods like semantic correlation maximization (SCM) [16,17] continuously learn hash codes by semantic label similarity. The semantic gaps are narrowed by making a probabilistic affinity matrix and minimizing the KL-divergences in the semantic preserving hashing (SePH) technique [5]. However, because of memory overhead and increased computational complexity, SePH can't handle large-scale datasets.…”

Section: Related Workmentioning

confidence: 99%

“…Dictionary learning attempts to learn the most naive feature as sparse representation to reserve the information comprehensively. Owing to the ability to extract essential data features into a sparse representation; Recently, numerous strategies based on dictionary learning [4][5][6] have been put out in the literature. However, technical hurdles exist to provide a more accurate and efficient hashing-based cross-modal retrieval by leveraging dictionary learning.…”

Section: Introductionmentioning

confidence: 99%

Label-consistent Kernel Transform Learning based Sparse Hashing for cross-modal retrieval

Maggu

2023

Preprint

View full text Add to dashboard Cite

Currently, there are various widely used methods for hashing-based cross modal retrieval to achieve promising performance in the information retrieval task. However, the challenge of learning discriminative representations against a heterogeneous gap in the case of multi-modal data still needs to be explored. Most existing methods directly project raw heterogeneous data linearly onto a common projection space, leading to high projection error because the semantic similarity of multi-modal data is not captured. We need the efficient feature representation of heterogeneous and multi-modal data to learn high-quality discrete hash codes and improve retrieval performance. This paper proposes a label-consistent kernel transform learning-based sparse hashing (LcKTLSH) for cross modal retrieval. Instead of projecting raw data samples, LcKTLSH learns transform and sparse coefficients suited for projection onto the kernel space, assuming each modality of a single instance has a unique hash code. The learned sparse coefficients are embedded into pre-learned hash codes, and semantic information from the similarity matrix is embedded into pre-learned hash codes to generate final hash codes. Kernel trick accounts for non-linearity in the data, and label-consistency constraint acts as the supervision term. Extensive experimental results on four benchmark datasets significantly advance over previous investigations.

show abstract

“…Combining these strengths holds significant potential. For instance, hybrid quantum-classical convolutional neural networks have demonstrated better performance compared to classical networks with the same architecture [14,15]. This is because generalized feature maps with variational quantum circuits can explore correlations between adjacent data points in exponentially large linear spaces, potentially allowing our algorithms to more accurately capture patterns within datasets using quantum computers.…”

Section: Introductionmentioning

confidence: 99%

HQC-MCDCNN: a novel hybrid quantum–classical multi-path denoising convolutional neural network

Fu,

Che,

Nie

et al. 2024

Phys. Scr.

View full text Add to dashboard Cite

Image denoising is a longstanding and enduring visual problem, and with the continuous rise of quantum computing in the field of machine learning, its role in image processing has become increasingly important. This paper introduces for the first time the use of multiscale variational quantum circuits in the field of image denoising, aiming to enhance the performance of classical convolutional neural networks and explore the potential advantages of combining quantum and classical approaches. In this work, we propose a novel Hybrid Quantum-Classical Multi-Path Denoising Convolutional Neural Network, abbreviated as HQC-MCDCNN. The HQC-MCDCNN is composed of a hybrid of quantum and classical elements, with the quantum part using multiscale variational quantum circuits instead of classical convolutional layers for feature extraction, and the classical part employing a newly designed multi-path denoising convolutional neural network for supervised learning. Together, these components synergistically achieve image denoising. It is worth noting that this paper aims to build readers' intuition for quantum computing, presenting all internal details of this work with rich images and visualizations. To demonstrate the denoising capability of HQC-MCDCNN, we conducted rigorous comparative experiments. Due to the constraints of Noisy Intermediate-Scale Quantum (NISQ) devices and the limited number of quantum bits, the experiments were based on the MNIST and Fashion-MNIST datasets with varying degrees of noise (noise factors ranging from 0.3 to 0.7), employing a 6-fold stratified sampling strategy for cross-validation. The experimental results indicate that HQC-MCDCNN is promising across all evaluation metrics, particularly outperforming other models by 56.5\% in the average UIQ index. This suggests that our hybrid model exhibits outstanding feature extraction capabilities and excellent denoising performance, providing a promising path for addressing image denoising challenges.

show abstract

Discriminative Fisher Embedding Dictionary Transfer Learning for Object Recognition

Cited by 13 publications

References 50 publications

Enhancing Image Annotation With Object Tracking and Image Retrieval: A Systematic Review

Enhancing Image Annotation With Object Tracking and Image Retrieval: A Systematic Review

Label-consistent Kernel Transform Learning based Sparse Hashing for cross-modal retrieval

HQC-MCDCNN: a novel hybrid quantum–classical multi-path denoising convolutional neural network

Contact Info

Product

Resources

About