Trademark Retrieval Research Articles

Trademark Similarity Evaluation Using a Combination of ViT and Local Features

The origin of the trademark similarity analysis problem lies within the legal area, specifically the protection of intellectual property. One of the possible technical solutions for this issue is the trademark similarity evaluation pipeline based on the content-based image retrieval approach. CNN-based off-the-shelf features have shown themselves as a good baseline for trademark retrieval. However, in recent years, the computer vision area has been transitioning from CNNs to a new architecture, namely, Vision Transformer. In this paper, we investigate the performance of off-the-shelf features extracted with vision transformers and explore the effects of pre-, post-processing, and pre-training on big datasets. We propose the enhancement of the trademark similarity evaluation pipeline by joint usage of global and local features, which leverages the best aspects of both approaches. Experimental results on the METU Trademark Dataset show that off-the-shelf features extracted with ViT-based models outperform off-the-shelf features from CNN-based models. The proposed method achieves a mAP value of 31.23, surpassing previous state-of-the-art results. We assume that the usage of an enhanced trademark similarity evaluation pipeline allows for the improvement of the protection of intellectual property with the help of artificial intelligence methods. Moreover, this approach enables one to identify cases of unfair use of such data and form an evidence base for litigation.

Learning test-time augmentation for content-based image retrieval

Off-the-shelf convolutional neural network features achieve outstanding results in many image retrieval tasks. However, their invariance to target data is pre-defined by the network architecture and training data. Existing image retrieval approaches require fine-tuning or modification of pre-trained networks to adapt to variations unique to the target data. In contrast, our method enhances the invariance of off-the-shelf features by aggregating features extracted from images augmented at test-time, with augmentations guided by a policy learned through reinforcement learning. The learned policy assigns different magnitudes and weights to the selected transformations, which are selected from a list of image transformations. Policies are evaluated using a metric learning protocol to learn the optimal policy. The model converges quickly and the cost of each policy iteration is minimal as we propose an off-line caching technique to greatly reduce the computational cost of extracting features from augmented images. Experimental results on large trademark retrieval (METU trademark dataset) and landmark retrieval (ROxford5k and RParis6k scene datasets) tasks show that the learned ensemble of transformations is highly effective for improving performance, and is practical, and transferable.

Unsupervised Trademark Retrieval Method Based on Attention Mechanism.

Aiming at the high cost of data labeling and ignoring the internal relevance of features in existing trademark retrieval methods, this paper proposes an unsupervised trademark retrieval method based on attention mechanism. In the proposed method, the instance discrimination framework is adopted and a lightweight attention mechanism is introduced to allocate a more reasonable learning weight to key features. With an unsupervised way, this proposed method can obtain good feature representation of trademarks and improve the performance of trademark retrieval. Extensive comparative experiments on the METU trademark dataset are conducted. The experimental results show that the proposed method is significantly better than traditional trademark retrieval methods and most existing supervised learning methods. The proposed method obtained a smaller value of NAR (Normalized Average Rank) at 0.051, which verifies the effectiveness of the proposed method in trademark retrieval.

Efficient retrieval of trademark images from large databases

Trademarks law is considered as the most pervasive amongst all the intellectual property laws because all the judgments require contemplating the imaginations of the consumers. Once the trademark has been submitted for registration, the examiners at the trademark office make sure that it is not similar to any of the previous registered trademarks. This motivated the need of an automated trademark retrieval system. This paper makes a contribution in the field of trademark image retrieval by proposing a retrieval technique that allows a flexible combination of colour, texture and shape features. Moreover the proposed technique utilises HSV colour histogram, for colour, multi resolution Gabor wavelet for texture and an integration of Zernike moments for global shape and scale invariant feature transform (SIFT) for local shape feature extraction. The results have been tested on MPEG7, MPEG trademark, WANG and self-compiled datasets. The improvement achieved in precision is 14% on MPEG7, 30% on the WANG and 38% on self-compiled dataset. Similarly, 26% improvement in average recall is achieved and around 16% when our proposed shape feature is compared with other state of the art techniques like Fourier descriptors, Hu moments, wavelet descriptors, Zernike moments and edge gradient co-occurrence matrix.

Component-Based Attention for Large-Scale Trademark Retrieval

The demand for large-scale trademark retrieval (TR) systems has significantly increased to combat the rise in international trademark infringement. Unfortunately, the ranking accuracy of current approaches using either hand-crafted or pre-trained deep convolution neural network (DCNN) features is inadequate for large-scale deployments. We show in this paper that the ranking accuracy of TR systems can be significantly improved by incorporating hard and soft attention mechanisms, which direct attention to critical information such as figurative elements and reduce attention given to distracting and uninformative elements such as text and background. Our proposed approach achieves state-of-the-art results on a challenging large-scale trademark dataset.

Indexing Techniques in Trademark Image Retrieval Systems

The development of automatic trademark image retrieval systems becomes a necessity because of the increasing number of registered trademarks in all countries. The goal is to protect the registered trademarks from counterfeiting and infringement. This paper introduces a trademark image retrieval system using indexing techniques. The proposed system is described by giving an overview about its architecture and describing in details all its components. The goal is to allow researchers and developers in image retrieval to build their own trademark retrieval system using the indexing techniques. Each part of the proposed system is considered as a component that can be improved or replaced. The reader can have a clear idea on: (1) the type of visual features to extract from the trademark images, (2) the indexing technique that can be used to organize the extracted features and speed-up the search and (3) how to perform a similar search for a new trademark image. The proposed system has been evaluated using several global features and the best performance is obtained when using Zernike moments coefficients with order 12.

Trademark Image Retrieval Based on Faster R-CNN

Automatic retrieval of digital trademark images is significant for improving the efficiency of trademark examination and management. In this paper, we proposed a method based on deep learning for trademark retrieval. Faster R-CNN was first applied to trademark retrieval. The global feature descriptor of the image is extracted by Faster R-CNN and the local feature of the image is extracted through the object proposal regions by the Region Proposal Networks (RPN). Retrieval strategies consist of initial ranking and spatial reranking. The experimental results show that our proposed method achieves remarkable performance in trademark image retrieval.

A Novel Shape Representation Method for Complex Trademark Image

Trademark orthogonal representation is an important research field in artificial intelligence technology and has extensive applications in trademark retrieval and recognition. However, the Gibbs phenomenon arises when complex trademark shapes are represented by using traditional methods, such as continuous orthogonal function and wavelet. In order to represent the complex trademark without Gibbs phenomenon, a novel method named hierarchical V-system (HV system) is proposed in this study, which is generated by V-system of multi-degree $k$ ( $k = 0, 1, 2, 3$ ). The hierarchical structure brings more detailed shape representation information. We show that the proposed method in this paper can represent the trademark shapes with fewer finite terms and obtain accurate representation result. This study also proposes the normalized descriptors of hierarchical V-system (HV descriptors). Furthermore, we demonstrate that the HV descriptors satisfy the invariance in rotation, translation, and scale transform. The experimental results show that the hierarchical V-system method can give reasonable descriptors for representing complex trademark without Gibbs phenomenon.

Deep CNN Combined With Relevance Feedback for Trademark Image Retrieval

Abstract Trademark recognition and retrieval is a vital appliance component of content-based image retrieval (CBIR). Reduction in the semantic gap, attaining more accuracy, reduction in computation complexity, and hence in execution time, are the major challenges in designing and developing a trademark retrieval system. The direction of the proposed work takes into account these challenges by implementing trademark image retrieval through deep convolutional neural networks (DCNNs) integrated with a relevant feedback mechanism. The dataset features are optimized through particle swarm optimization (PSO), reducing the search space. These best/optimized features are given to the self-organizing map (SOM) for clustering at the preprocessing stage. The CNN model is trained on feature representations of relevant and irrelevant images, using the feedback information from the user bringing the marked relevant images closer to the query. Experimentation proved a significant performance when evaluated using FlickrLogos-27, FlickrLogos-32, and FlickrLogos-32 PLUS datasets, as illustrated in the performance results section.

Translated Trademarks Retrieval using Color Autocorrelogram for Extracted Textual Parts

This article describes a trademark retrieval system using autocorrelogram features. The proposed algorithm deals with the trademarks of a same company having different shapes in different languages. Textual and mixed logos types are considered in this study, in which the text area is located based on an edge detection algorithm and connected components merging strategy to form the bounding box enclosing the textual parts of a logo. Color autocorrelogram is applied on the extracted and quantified text area. An adapted evaluation metric is proposed to measure the performance of the proposed technique. Results show a good retrieval rate on real trademark images.

Novel Relevance Feedback Approach for Color Trademark Recognition Using Optimization and Learning Strategy

Abstract The trademark registration process, apparent in all organizations nowadays, deals with recognition and retrieval of similar trademark images from trademark databases. Trademark retrieval is an imperative application area of content-based image retrieval. The main challenges in designing and developing this application area are reducing the semantic gap, obtaining higher accuracy, reducing computation complexity, and subsequently the execution time. The proposed work focuses on these challenges. This paper proposes the relevance feedback system embedded with optimization and unsupervised learning technique as the preprocessing stage, for trademark recognition. The search space is reduced by using particle swam optimization, for optimization of database feature set, which is further followed by clustering using self-organizing map. The relevance feedback technique is implemented over this preprocessed feature set. Experimentation is done using the FlickrLogos-32 PLUS dataset. To introduce variations between the training and query images, transformations are applied to each of the query image, viz. rotation, scaling, and translation of the image. The same query image is tested for various combinations of transformations. The proposed technique is invariant to various transformations, with significant performance as depicted in the results.

Semantic Retrieval of Trademarks Based on Conceptual Similarity

Trademarks are signs of high reputational value. Thus, they require protection. This paper studies conceptual similarities between trademarks, which occurs when two or more trademarks evoke identical or analogous semantic content. This paper advances the state-of-the-art by proposing a computational approach based on semantics that can be used to compare trademarks for conceptual similarity. A trademark retrieval algorithm is developed that employs natural language processing techniques and an external knowledge source in the form of a lexical ontology. The search and indexing technique developed uses similarity distance, which is derived using Tversky’s theory of similarity. The proposed retrieval algorithm is validated using two resources: a trademark database of 1400 disputed cases and a database of 378 943 company names. The accuracy of the algorithm is estimated using measures from two different domains: the ${R}$ -precision score, which is commonly used in information retrieval and human judgment/collective human opinion, which is used in human–machine systems.

FUZZY CLUSTERING OF IMAGE TRADEMARK DATABASE AND PREPROCESSING USING ADAPTIVE FILTER AND KARHUNEN–LOÈVE TRANSFORM

In this paper, an efficient preprocessing module has been described which focuses on building a trademark database that can be used for developing a trademark retrieval system. The preprocessing module focuses on noise removal from the trademark images using an adaptive filtering technique using Wiener filters, followed by Karhunen–Loève transform that makes the trademark search process rotation invariant by rotating the object along positive y direction. Since the registered trademarks are huge in number and will increase invariantly in the future it will be strenuous for the search system to search for similarity in such huge database. Intention was to reduce the search space hence fuzzy clustering has been applied. All these preprocessing steps make a retrieval system more efficient and reduce computation cost.

A Conceptual Model of Trademark Retrieval based on Conceptual Similarity

The rapid expansion of e-commerce at the beginning of 21st century has had a significant impact on intellectual property management. A particular area of concern is the misuse of trademarks and trademark protection. Trademarks are proprietary words and images with high reputational value; they are important assets, often used as a marketing tool, which require infringement protection. One of the issues considered during infringement litigation is the visual, conceptual and phonetic similarity of different trademarks. In particular, the conceptual similarity of trademarks is an area never previously studied in information retrieval. This paper focuses on this important aspect by proposing a conceptual model of the comparison process, aimed at retrieving conceptually similar trademarks. The proposed model employs natural language processing and semantic technology to compute the conceptual similarity between trademarks.

An Affinity Based Complex Artificial Immune System and its Application on Trademark Retrieval

Recently, artificial immune system (AIS) inspired by the theory or immunology, has been developed rapidly and steadily. In this paper, we proposed an affinity based complex artificial immune system model considering the fact that the different eptitopes located on the surface of antigen can be recognized by a set of different paratopes expressed on the surface of immune cells. The experiment on trademark retrieval is performed to prove that the model proposed model has an excellent performance on retrieving the trademark images and outperforms the previously proposed algorithm.

A NOVEL ALGORITHM FOR TRADEMARK RETRIEVAL BASED ON V-SYSTEM

Trademark image retrieval is one of important means for trademark distinctness check;the purpose is to review the repeatability and similarity among trademark images.This paper proposes a new trademark retrieval method based on an orthogonal function system called V-system.The contour of a trademark is accurately expressed with finite basis orthogonal functions of V-system,and then its eigenvectors are extracted.The Euclidean distance between eigenvectors is used to calculate similarity of two trademarks,and then trademark retrieval can be carried out.We compare our method with Fourier descriptor and Zernike moments, and finally,the experiment results indicate that V-system has great advantage on trademark retrieval.

Trademark image retrieval using an integrated shape descriptor

Trademarks are distinctive visual symbols with high reputational value, due to the perception of quality and innovation associated with them. They are important reputational assets used as a marketing tool to convey a certain assurance of quality, innovation, and the standards, which the manufacturer seeks to maintain. This motivates the need for trademark protection by providing a solution to prevent infringement. This problem can be addressed by developing retrieval systems capable of comparing the visual similarity of trademarks. This paper contributes to the research in this field by proposing an innovative trademark retrieval technique with improved retrieval performance due to the integration of global and local descriptors. The global descriptor employed is the Zernike moment’s coefficients. The local descriptor is the edge-gradient co-occurrence matrix, derived from the contour information that is considered very important in human perception of visual similarity. The proposed retrieval technique is tested using the standard MPEG-7 shape database of 1400 images and the MPEG-7 trademark database of 3260 images. The results show 5% precision/recall improvement in the case of the MPEG-7 shape database, as well as 2.35% Bull’s eye score improvement and 19.8% NMRR score improvement for the 10 randomly selected trademarks from the MPEG-7 trademarks database.

An Effective Solution for Trademark Retrieval by Combining Pseudo Zernike Moments and SIFT Features

This paper proposes a trademark retrieval algorithm combining the image global features and local features. Firstly, extract pseudo Zernike moments of the retrieved image and sort them according to similarity. Candidate images are formed. Then, the SIFT features are used for matching the query image accurately with candidate images. Experimental results show that this method not only keeps high precision- recall of SIFT features and is superior than the method based on the single pseudo Zernike moments feature, but also improves effective retrieval speed compared to the single SIFT features. This method can be well applied to the trademark image retrieval system.

Design of Online System for Trademark Retrieval Based on Composite Features

Design of Online System for Trademark Retrieval Based on Composite Features

Content based Trademark Retrieval by Integrating Shape with Colour and Texture Information

is one of the most disputed items in Intellectual Property Rights. To grant a trademark, litigation is carried out to investigate if the new trademark applied for matches substantially with an already issued trademark. Considering the fact that the trademark database is very huge and keeps on increasing constantly, the manual search is definitely not a correct solution and an Information and Communication Technology (ICT) based system is required to automate this process. In a content based retrieval system, shape is easily the most important feature for the trademark image. Using shape feature various systems have been developed in the past. Here, a new approach is proposed that integrates shape with color and texture information that improves the results significantly in terms of Precision and Recall. A flexible combination of features is used for improving performance and clustering is applied to the database to enhance retrieval efficiency.