T R A C K       P A P E R

Login Panel

Close tab

Password Meter

Volume 10 Issue 11 (November 2023)

S.No. Title & Authors Page No View

Title : Research on Vehicle Object Detection Algorithm Based on Deep Learning

Authors : Zhengkun Shen, Huaixian Yin

Click Here For Abstract

Download Certificate
Abstract :

Aiming at the low detection efficiency and accuracy of existing deep learning vehicle target detection algorithms, a VOD-YOLOv5 vehicle target detection algorithm was proposed based on the YOLOv5 model. In this paper, a new lightweight convolutional neural module (cf) is proposed to improve the feature extraction capability of the network. At the same time, the attention mechanism of space and channel fusion is integrated into the network model, which improves the detection accuracy of small and medium-sized targets in fuzzy images. The experimental results show that compared with the original YOLOv5 model, the VOD-YOLOv5 model proposed in this paper has a 4% increase in average accuracy (mAP), and the average accuracy (AP) of detection of different target classes has been improved, and the detection speed meets the real-time requirements, effectively improving the detection performance of the vehicle target detection model.


Title : Directed Acyclic Graphs using Approximate Labels for Conversational Emotion Recognition

Authors : Wenxue Zhang, Baoshan Sun

Click Here For Abstract

Download Certificate
Abstract :

Conversational context modelling is a crucial aspect that holds significance in emotion recognition from conversation. In this paper, we present the directed acyclic graph using approximate labels model(LDAG-ERC), which aims to enhance the performance of the current ERC dialogue task model, DAG-ERC.The LDAG-ERC model is a modified version of the DAG-ERC model. It uses directed acyclic graphs (DAG) with approximate emotion labels to describe the internal structure of the dialogue. Our LDAG-ERC model has significantly improved performance compared to the DAG-ERC model. We conducted experiments on an ERC datasets and compared them to the original model to illustrate the effectiveness of the modifications.