The PASCAL Visual Object Classes Challenge

We introduce a purely feed-forward architecture for semantic segmentation. We map small image elements (superpixels) to rich feature representations extracted from a sequence of nested regions of increasing extent. These regions are obtained by “zooming out” from the superpixel all the way to scene-level resolution. This approach exploits statistical structure in the image and in the label space without setting up explicit structured prediction mechanisms, and thus avoids complex and expensive inference. Instead superpixels are classified by a feedforward multilayer network. Our architecture achieves 69.6% average accuracy on the PASCAL VOC 2012 test set.

/pdf/feedforward-semantic-segmentation-with-zoom-out-features-256wutztcc.pdf

Feedforward semantic segmentation with zoom-out features

Automatic License Plate Recognition (ALPR) has been a frequent topic of research due to many practical applications. However, many of the current solutions are still not robust in real-world situations, commonly depending on many constraints. This paper presents a robust and efficient ALPR system based on the state-of-the-art YOLO object detector. The Convolutional Neural Networks (CNNs) are trained and fine-tuned for each ALPR stage so that they are robust under different conditions (e.g., variations in camera, lighting, and background). Specially for character segmentation and recognition, we design a two-stage approach employing simple data augmentation tricks such as inverted License Plates (LPs) and flipped characters. The resulting ALPR approach achieved impressive results in two datasets. First, in the SSIG dataset, composed of 2,000 frames from 101 vehicle videos, our system achieved a recognition rate of 93.53% and 47 Frames Per Second (FPS), performing better than both Sighthound and OpenALPR commercial systems (89.80% and 93.03%, respectively) and considerably outperforming previous results (81.80%). Second, targeting a more realistic scenario, we introduce a larger public dataset, called UFPR-ALPR dataset, designed to ALPR. This dataset contains 150 videos and 4,500 frames captured when both camera and vehicles are moving and also contains different types of vehicles (cars, motorcycles, buses and trucks). In our proposed dataset, the trial versions of commercial systems achieved recognition rates below 70%. On the other hand, our system performed better, with recognition rate of 78.33% and 35 FPS.

A Robust Real-Time Automatic License Plate Recognition based on the YOLO Detector

Automatic License Plate Recognition (ALPR) has been a frequent topic of research due to many practical applications. However, many of the current solutions are still not robust in real-world situations, commonly depending on many constraints. This paper presents a robust and efficient ALPR system based on the state-of-the-art YOLO object detector. The Convolutional Neural Networks (CNNs) are trained and finetuned for each ALPR stage so that they are robust under different conditions (e.g., variations in camera, lighting, and background). Specially for character segmentation and recognition, we design a two-stage approach employing simple data augmentation tricks such as inverted License Plates (LPs) and flipped characters. The resulting ALPR approach achieved impressive results in two datasets. First, in the SSIG dataset, composed of 2,000 frames from 101 vehicle videos, our system achieved a recognition rate of 93.53% and 47 Frames Per Second (FPS), performing better than both Sighthound and OpenALPR commercial systems (89.80% and 93.03%, respectively) and considerably outperforming previous results (81.80%). Second, targeting a more realistic scenario, we introduce a larger public dataset1 dataset, designed to ALPR. This dataset contains 150 videos and 4,500 frames captured when both camera and vehicles are moving and also contains different types of vehicles (cars, motorcycles, buses and trucks). In our proposed dataset, the trial versions of commercial systems achieved recognition rates below 70%. On the other hand, our system performed better, with recognition rate of 78.33% and 35 FPS.The UFPR-ALPR dataset is publicly available to the research community at https://web.inf.ufpr.br/vri/databases/ufpr-alpr/ subject to privacy restrictions.

A Robust Real-Time Automatic License Plate Recognition Based on the YOLO Detector

Viable cryptosystem designs must address power analysis attacks, and masking is a commonly proposed technique for defending against these side-channel attacks It is possible to overcome simple masking by using higher-order techniques, but apparently only at some cost in terms of generality, number of required samples from the device being attacked, and computational complexity We make progress towards ascertaining the significance of these costs by exploring a couple of attacks that attempt to efficiently employ second-order techniques to overcome masking In particular, we consider two variants of second-order differential power analysis: Zero-Offset 2DPA and FFT 2DPA

Towards efficient second-order power analysis

This paper presents an online highly accurate system for automatic number plate recognition (ANPR) that can be used as a basis for many real-world ITS applications. The system is designed to deal with unclear vehicle plates, variations in weather and lighting conditions, different traffic situations, and high-speed vehicles. This paper addresses various issues by presenting proper hardware platforms along with real-time, robust, and innovative algorithms. We have collected huge and highly inclusive data sets of Persian license plates for evaluations, comparisons, and improvement of various involved algorithms. The data sets include images that were captured from crossroads, streets, and highways, in day and night, various weather conditions, and different plate clarities. Over these data sets, our system achieves 98.7%, 99.2%, and 97.6% accuracies for plate detection, character segmentation, and plate recognition, respectively. The false alarm rate in plate detection is less than 0.5%. The overall accuracy on the dirty plates portion of our data sets is 91.4%. Our ANPR system has been installed in several locations and has been tested extensively for more than a year. The proposed algorithms for each part of the system are highly robust to lighting changes, size variations, plate clarity, and plate skewness. The system is also independent of the number of plates in captured images. This system has been also tested on three other Iranian data sets and has achieved 100% accuracy in both detection and recognition parts. To show that our ANPR is not language dependent, we have tested our system on available English plates data set and achieved 97% overall accuracy.

Accurate Detection and Recognition of Dirty Vehicle Plate Numbers for High-Speed Applications

In visual surveillance, detecting and localising abnormal events are of great interest. In this study, an unsupervised method is proposed to automatically discover abnormal events occurring in traffic videos. For learning typical motion patterns occurring in such videos, a group sparse topical coding (GSTC) framework and an improved version of it are applied to optical flow features extracted from video clips. Then a very simple and efficient algorithm is proposed for GSTC. It is shown that discovered motion patterns can be employed directly in detecting abnormal events. A variety of abnormality metrics based on the resulting sparse codes for detection of abnormality are investigated. Experiments show that the result of the approach in detection and localisation of abnormal events is promising. In comparison with other usual methods (probabilistic latent semantic analysis, latent Dirichlet allocation, sparse topical coding (STC) and improved STC), according to the values of area under ROC, the proposed method achieves at least 14% improvement in abnormal event detection.

https://ietresearch.onlinelibrary.wiley.com/doi/epdf/10.1049/iet-ipr.2015.0399

Abnormal event detection and localisation in traffic videos based on group sparse topical coding

Automatic analysis and understanding of typical activities and identification of abnormal events in crowded traffic scenes is a fundamental task for traffic video surveillance. In this paper, we address the problem of abnormality detection based on an unsupervised learning approach with Fully Sparse Topic Models (FSTM). The method uses a set of visual features and automatically discovers the activity patterns occurring in complicated scenes. We show how the discovered patterns can be used to detect abnormal events. Furthermore, we compare FSTM with other topic models based on various measures. Experimental results and comparisons on two traffic datasets demonstrate that our approach outperforms other methods in finding meaningful activity patterns and discovers the abnormal events accurately.

Incorporating fully sparse topic models for abnormality detection in traffic videos

With the extensive growth in information, text classification as one of the text mining methods, plays a vital role in organizing and management information. Most text classification methods represent a documents collection as a Bag of Words (BOW) model and then use the histogram of words as the classification features. But in this way, the number of features is very large; therefore performing text classification faces serious computational cost problems. Moreover, the BOW representation is unable to recognize semantic relations between words. Recently, topic-model approaches have been successfully applied for text classification to overcome the problems of BOW. Our main goal in this paper is to investigate the possibility of applying the topic models for Persian text classification and compare between the feature processing techniques of BOW and the topic model based approaches. The experimental results show that the topic-model approach for representing the Persian documents yields at least 9% accuracy improvement compared to the BOW based algorithm.

Persian text classification based on topic models

Automatic accident detection is one of the most important tasks for an intelligent transportation system (ITS). In this paper, a new framework for automated traffic accident recognition using topic models is proposed. This framework uses a set of visual features and automatically discovers the motion patterns in traffic scenes. Then, using these learned motion patterns, occurrence of an accident could be detected by various abnormality measures. Results on real video sequences collected from Tehran traffic control center confirm the effectiveness and the applicability of the proposed framework.

Iman Gholampour

Papers

Accurate Detection and Recognition of Dirty Vehicle Plate Numbers for High-Speed Applications

Abnormal event detection and localisation in traffic videos based on group sparse topical coding

Incorporating fully sparse topic models for abnormality detection in traffic videos

Persian text classification based on topic models

Automatic accident detection using topic models