Pattern Recognition is the official journal of the Pattern Recognition Society. The Society was formed to fill a need for information exchange among research workers in the pattern recognition field. Up to now, we ''pattern-recognitionophiles'' have been tagging along in computer science, information theory, optical processing techniques, and other miscellaneous fields. Because this work in pattern recognition presently appears in widely spread articles and as isolated lectures in conferences in many diverse areas, the purpose of the journal Pattern Recognition is to give all of us an opportunity to get together in one place to publish our work. The journal will thereby expedite communication among research scientists interested in pattern recognition.

We consider pattern recognition in the broad sense, and we assume that the journal will be read by people with a common interest in pattern recognition but from many diverse backgrounds. These include biometrics, target recognition, biological taxonomy, meteorology, space science, classification methods, character recognition, image processing, industrial applications, neural computing, and many others.

The publication policy is to publish (1) new original articles that have been appropriately reviewed by competent scientific people, (2) reviews of developments in the field, and (3) pedagogical papers covering specific areas of interest in pattern recognition. Various special issues will be organized from time to time on current topics of interest to Pattern Recognition.

Special Issue on Weakly Supervised Learning on Big Vision Data
Submission Date: 2019-10-15

Recently, pattern recognition and computer vision have been largely boosted by the use of deep learning. This boost has been seen in a wide range of applications, including object detection, tracking and recognition, action recognition, face recognition, person re-identification, etc. Such a huge development is due to advancing hardware and also the availability of big data (e.g. the large number of data samples).

Actually, there are large demands in many computer vision tasks, such as object detection and untrimmed action recognition. It is also costly to annotate the location of each object/action in an image/video for training a large-scale set of images/videos. And thus, it is well demanded to seek a way to minimize such a requirement.

More importantly, while a huge number of data samples make deep neural networks learnable, most of the developed algorithms largely depend on labelled data, acquiring labels is extremely expensive for increasingly big data, and it is ultimately infeasible to label all data samples. How to utilize partially labelled and even unlabelled data to learn feature representation by deep neural networks is now critically important. Recently, several attempts have been made, including the extension of dictionary learning and clustering-based feature extraction under deep neural networks for achieving an end-to-end learning recently as well. The multi-instance learning is employed since it is suitable for weakly labelled data, for instance when providing the existence label of an object in an image without localization. The curriculum learning that learns the knowledge from the easy samples (e.g. salient candidate in object detection) to hard samples is also developed to solve the weakly supervised learning challenge. The related approach such as distillation methods/teacher-student models are also found useful to help transfer existing strong network to the one with the same structure trained on only limited data. More recently, the developed deep adversarial network and deep capsule network are becoming important approaches for unsupervised learning, providing a promising way for using unlabeled data in deep neural networks effectively.

In addition, while the large applications of big data by deep neural network are for classification problem (e.g. object detection/categorization, action recognition), a more important application of big data is to forecast what will happen or what the trend is in future periods based on the previous observed data and prior knowledge. Unsupervised learning is important for predictive learning, since prediction is based on a vast amount of historical data, which are apparently impossible to be completely labelled and therefore mostly are unlabelled.

We call the setting of learning under limited labelled data or limited annotation as weakly supervised learning. We are still at the starting points of solving the weakly supervised learning. This special issue aims to promote the research on the purpose of weakly supervised learning on big data. We aim to solicit high quality papers, both in theory and applications. Especially, the following (but not limited to) topics are the particular interests of this special issue, including:

Unsupervised Learning
Semi-supervised Learning
Deep Clustering
Adversarial Learning
Predictive Learning
Distillation Modelling
Online Learning
Active Learning
Transfer Learning
Related applications in computer vision and pattern recognition

Special Issue on Pattern Recognition Methods for Bridging the Vision-Language Gap in Multimodal Data Analysis
Submission Date: 2020-02-12

The explosive growth of visual and textual data (both on the WorldWideWeb and held in private repositories by diverse institutions and companies) has led to urgent requirements in terms of searching, processing and understanding of multimedia content, by a machine. Solutions for providing access to and understanding such multimodal source data depend on bridging the semantic gap between vision and language. To solve this problem calls for expertise from the cognate fields of computer vision, image processing, text and document analysis, machine learning and pattern recognition. This problem also finds applications in the fast-emerging areas of multimedia data analysis and cross-modality learning.

In this special issue, we aim to assemble recent advances in pattern recognition relevant to the vision-and-language problem, encompassing big-data applications involving multimedia data and deep learning algorithms. The scope of the call for papers covers the use of pattern recognition and machine learning techniques for understanding cross-modal information, especially to those involving vision-and-language. Both original research as well as state-of-the-art literature reviews, are welcome for submission. However, submitted papers must be within the scope of the Pattern Recognition Journal, advancing the available pattern recognition methodology in this domain. Papers outside the remit of the journal will be rejected without review. The list of possible topics includes, but is not limited to:

Novel pattern recognition and machine learning methods which combine language and vision
Pattern recognition and machine learning for visual captioning, dialogue, and question answering
Sequence learning towards bridging vision, language and multimedia data
Language as an inference mechanism for structuring and reasoning about visual perception
Transfer learning across multimodal data
Pattern recognition for visual synthesis from language
Semantic scene graph generation from images with pattern recognition and machine learning methods
Cross-modality pattern recognition and machine learning for representation and learning, retrieval and generation, and zero/few-shot learning.
Pattern recognition and machine learning for multimedia data analysis and understanding

Special Issue on Modeling and Learning for Matching: Models, Methods and Applications
Submission Date: 2020-12-30

The special issue will focus on the recent advance in modeling and learning to solve the matching problem in pattern recognition. The capability of finding correspondence among images, graphics, point sets and other structures has been a fundamental in many pattern recognition. The past decades have witnessed the rapid expansion of the frontier for automatic correspondence establishment among images/graphics, which is largely due to the advances in computational capacity, data availability and new algorithmic paradigms. Although the correspondence problem has been extensively studied in the context of multi-view geometry, its more generalized forms, along with underlying connections among different methods and settings, have not been fully explored. Meanwhile, the combination of big data and the deep learning paradigm has achieved significant success in many perceptual tasks; however, the existing paradigm is still far from a panacea to the matching problem, which often calls for more careful treatments on the local and global structures. Also, there are emerging methods for discovering latent graph structures that enrich the applicability for graph matching. In this special issue, we attempt to assemble recent advances in the matching problem, considering the explosions of big visual data applications and the deep learning algorithms.

This special issue will feature original research papers related to the models and algorithms for robust establishment of correspondence, together with applications to real-world problems.

Main Topics of Interest (but are not limited to):

Object matching: 1) Graph representation and modeling using image/graphics data; 2) Robust matching/registration for visual correspondences over two or multiple images/graphics/point sets; 3) Partial, one-to-many/many-to-many matching, in the presence of major noise and outliers; 4) Similarity between graphs/graphics and graph clustering/classification; 5) Cross-network matching, e.g. social networks and other forms e.g. protein network; 6) Incremental matching of a series of objects; 7) Shape matching.

Tracking and optical flow estimation: 1) Single/multiple object tracking and data association; 2) Robust and/or efficient optical flow methods; 3) Object co-detection; 4) Visual trajectory generation and modeling; 5) Person Re-ID; 6) Planar Object Tracking

Correspondence for 3-D vision: 1) Calibration, and pose estimation; 2) visual SLAM; 3) Depth estimation and 3-D reconstruction;

Learning for/by permutation and matching: 1) Learning graph structure and similarity from data with established or unestablished correspondences; 2) Learning image feature representation from established or loosely established correspondence; 3) Common/similar objects discovery and recognition from images; 4) Learning for matching and permutation.

Structure discovery from data: 1) structure inference from behavior data e.g. time series and event sequence; 2) latent structure matching based on behavior data

Applications: Application of matching technology to solve any real-world visual understanding problems including object detection/recognition among images/graphics, image stitching, 2-D/3-D recovery, robot vision, photogrammetry and remote sensing, industrial imaging, embed system etc.