Search

Vehicle Exterior Body Damage(VEBD) parts classification is important for claim processing, cost estimation, accident investigation, and vehicle damage assessment. Class imbalance in the VEBD part classification dataset is a primary factor affecting the classification performance of existing classification models. Although the availability of datasets and the system’s capability significantly impact system performance, its damage part classification is still limited due to its dynamic body structure, size, shape, color, and types of damage. In this paper, we propose a novel heterogeneous ensemble learning (HEL) model based on VEBD data (VEBD-HEL) to deal with imbalanced data in VEBD. We validate the effectiveness of VEBD-HEL on two original and generated VEBD datasets. The experimental results demonstrate that compared with current state-of-the-art models, VEBD-HEL has the best comprehensive performance. The proposed model not only achieves good Accuracy (99.93%) rates for both the simple damage and the severe damage but also increases the Area Under Curve (AUC) to 99.83%.

In a conventional speech emotion recognition (SER) task, a classifier for a given language is trained on a pre-existing dataset for that same language. However, where training data for a language do not exist, data from other languages can be used instead. We experiment with cross-lingual and multilingual SER, working with Amharic, English, German, and Urdu. For Amharic, we use our own publicly available Amharic Speech Emotion Dataset (ASED). For English, German and Urdu, we use the existing RAVDESS, EMO-DB, and URDU datasets. We followed previous research in mapping labels for all of the datasets to just two classes: positive and negative. Thus, we can compare performance on different languages directly and combine languages for training and testing. In Experiment 1, monolingual SER trials were carried out using three classifiers, AlexNet, VGGE (a proposed variant of VGG), and ResNet50. The results, averaged for the three models, were very similar for ASED and RAVDESS, suggesting that Amharic and English SER are equally difficult. Similarly, German SER is more difficult, and Urdu SER is easier. In Experiment 2, we trained on one language and tested on another, in both directions for each of the following pairs: Amharic↔German, Amharic↔English, and Amharic↔Urdu. The results with Amharic as the target suggested that using English or German as the source gives the best result. In Experiment 3, we trained on several non-Amharic languages and then tested on Amharic. The best accuracy obtained was several percentage points greater than the best accuracy in Experiment 2, suggesting that a better result can be obtained when using two or three non-Amharic languages for training than when using just one non-Amharic language. Overall, the results suggest that cross-lingual and multilingual training can be an effective strategy for training an SER classifier when resources for a language are scarce.

In this paper, we create EMIR, the first-ever Music Information Retrieval dataset for Ethiopian music. EMIR is freely available for research purposes and contains 600 sample recordings of Orthodox Tewahedo chants, traditional Azmari songs and contemporary Ethiopian secular music. Each sample is classified by five expert judges into one of four well-known Ethiopian Kiñits, Tizita, Bati, Ambassel and Anchihoye. Each Kiñit uses its own pentatonic scale and also has its own stylistic characteristics. Thus, Kiñit classification needs to combine scale identification with genre recognition. After describing the dataset, we present the Ethio Kiñits Model (EKM), based on VGG, for classifying the EMIR clips. In Experiment 1, we investigated whether Filterbank, Mel-spectrogram, Chroma, or Mel-frequency Cepstral coefficient (MFCC) features work best for Kiñit classification using EKM. MFCC was found to be superior and was therefore adopted for Experiment 2, where the performance of EKM models using MFCC was compared using three different audio sample lengths. 3s length gave the best results. In Experiment 3, EKM and four existing models were compared on the EMIR dataset: AlexNet, ResNet50, VGG16 and LSTM. EKM was found to have the best accuracy (95.00%) as well as the fastest training time. However, the performance of VGG16 (93.00%) was found not to be significantly worse (P < 0.01). We hope this work will encourage others to explore Ethiopian music and to experiment with other models for Kiñit classification.

Detecting and classifying vehicles as objects from images and videos is challenging in appearance-based representation, yet plays a significant role in the substantial real-time applications of Intelligent Transportation Systems (ITSs). The rapid development of Deep Learning (DL) has resulted in the computer-vision community demanding efficient, robust, and outstanding services to be built in various fields. This paper covers a wide range of vehicle detection and classification approaches and the application of these in estimating traffic density, real-time targets, toll management and other areas using DL architectures. Moreover, the paper also presents a detailed analysis of DL techniques, benchmark datasets, and preliminaries. A survey of some vital detection and classification applications, namely, vehicle detection and classification and performance, is conducted, with a detailed investigation of the challenges faced. The paper also addresses the promising technological advancements of the last few years.

Teaching assistants (TAs) often lead courses using curricula they did not design. Therefore, examining how curriculum and professional development (PD) interact to influence TAs' teaching practices is critical. This study describes the effects of a curriculum and PD intervention in two contexts: when TAs are teaching curriculum that is explicitly linked to PD, and when teaching curriculum that is not linked to PD. The Intervention curriculum featured structured opportunities for reform-oriented teaching practices. The Intervention PD was situated in the context of these specific curriculum activities and modelled the desired teaching practices. TAs that participated in the intervention implemented more student-centered teaching practices than TAs that did not participate in the intervention, even when teaching curriculum that was not designed to be student-centered and was not linked to PD. A linear model of TAs' teaching practices that included PD type, task cognitive demand and curriculum type indicates that cognitive demand has the largest relationship with teaching practices, followed by PD type. These results have implications for policy. They suggest that investment in curriculum-linked TA PD can be effective even when teaching curricula that is not linked to PD. Additionally, investment in development of higher-cognitive-demand tasks may be an effective strategy to support implementation of student-centered practices.

ABSTRACT Biologists consider variability during biological investigations. A robust quantitative understanding of variability is particularly important during data analysis, where statistics are used to quantify variation and draw conclusions about phenomena while accounting for variation. Many students struggle to correctly apply a quantitative understanding of variation to statistically analyze data. We present quantitative and qualitative analyses of introductory biology students’ responses on two pairs of multiple-choice questions querying two concepts related to the quantitative analysis of variation. More students correctly identify a mathematical expression of variation than correctly interpret it. Many students correctly interpret a nonsignificant p-value in the context of a very small sample size, but fewer students do so in the context of a large sample size. These results imply that many students have an incomplete quantitative understanding of variation. These findings suggest that instruction focusing on conceptual understanding, not procedural problem solving, may elevate students’ quantitative understanding of variation.

Crack inspections of automotive engine components are usually conducted manually; this is often tedious, with a high degree of subjectivity and cost. Therefore, establishing a robust and efficient method will improve the accuracy and minimize the subjectivity of the inspection. This paper presents a robust approach towards crack classification, using transfer learning and fine-tuning to train a pre-trained ConvNet model. Two deep convolutional neural network (DCNN) approaches to training a crack classifier—namely, via (1) a Light ConvNet architecture from scratch, and (2) fined-tuned and transfer learning top layers of the ConvNet architectures of AlexNet, InceptionV3, and MobileNet—are investigated. Data augmentation was utilized to minimize over-fitting caused by an imbalanced and inadequate training sample. Data augmentation improved the accuracy index by 4%, 5%, 7%, and 4%, respectively, for the proposed four approaches. The transfer learning and fine-tuning approach achieved better recall and precision scores. The transfer learning approach using the fine-tuned features of MobileNet attained better classification accuracy and is thus proposed for the training of crack classifiers. Moreover, we employed an up-to-date YOLOv5s object detector with transfer learning to detect the crack region. We obtained a mean average precision (mAP) of 91.20% on the validation set, indicating that the model effectively distinguished diverse engine part cracks.