Enhanced Real-Time Object Detection using YOLOv7 and MobileNetv3

Résumé

Object detection serves as a crucial element in computer vision, increasingly relying on deep learning techniques. Among various methods, the YOLO series has gained recognition as an effective solution. This research enhances object detection by merging YOLOv7 with MobileNetv3, known for its efficiency and feature extraction. The integrated model was tested using the COCO dataset, which contains over 164,000 images across 80 categories, achieving a mAP score of 0.61. Additionally, confusion matrix analysis confirmed its accuracy, especially in detecting common objects such as 'person' and 'car' with minimal misclassifications. The results demonstrate the potential of the proposed model to address the complexities of real-world scenarios, highlighting its applicability in various scientific and industrial domains.

Citer ce document

Accès au document

Texte intégral en lecture en ligne, réservé aux abonnés SPHAERO et aux membres de l'institution. Se connecter

Auteur(s)

Statistiques

Consultations : 1

Téléchargements : 0