47325 research outputs found
Sort by
Christmas Card From Ella C. Wilson to Eartha White
Get PDFCorrespondence: Christmas greetings from Ella C. Wilson, 924 Duval, to Eartha White and everyone. Undated
UNF VS Embry-Riddle [Neg# 17A]
No full textRoll of B+W Film scanned, UNF VS Embry-Riddle, Undatedhttps://digitalcommons.unf.edu/spinnaker-images/1160/thumbnail.jp
UNF VS Embry-Riddle [Neg# 19A]
No full textRoll of B+W Film scanned, UNF VS Embry-Riddle, Undatedhttps://digitalcommons.unf.edu/spinnaker-images/1162/thumbnail.jp
UNF VS Embry-Riddle [Neg# 25A]
No full textRoll of B+W Film scanned, UNF VS Embry-Riddle, Undatedhttps://digitalcommons.unf.edu/spinnaker-images/1167/thumbnail.jp
UNF VS Embry-Riddle_Roll2 [Neg # 11]
No full textRoll of B+W Film scanned, UNF VS Embry-Riddle, Roll 2, Undatedhttps://digitalcommons.unf.edu/spinnaker-images/1188/thumbnail.jp
UNF VS Embry-Riddle_Roll2 [Neg # 17]
No full textRoll of B+W Film scanned, UNF VS Embry-Riddle, Roll 2, Undatedhttps://digitalcommons.unf.edu/spinnaker-images/1194/thumbnail.jp
UNF VS Embry-Riddle_Roll2 [Neg # 24]
No full textRoll of B+W Film scanned, UNF VS Embry-Riddle, Roll 2, Undatedhttps://digitalcommons.unf.edu/spinnaker-images/1201/thumbnail.jp
UNF VS Embry-Riddle_Roll2 [Neg # 25]
No full textRoll of B+W Film scanned, UNF VS Embry-Riddle, Roll 2, Undatedhttps://digitalcommons.unf.edu/spinnaker-images/1202/thumbnail.jp
Comparative analysis: Advanced model improvements from YOLOv8, YOLOv9 and YOLOv10 in traffic analysis
No full textThis thesis introduces a customized adaptation of YOLO object detection models (YOLOv8, YOLOv9, and YOLOv10) to address the distinct demands of urban traffic monitoring. Traditional YOLO applications often struggle with varied environmental conditions and high-density vehicle classes typical in real-world traffic data. By contrast, this research adapts YOLO models specifically for traffic analysis, implementing tailored pre-processing methods, adjusted input formats, and robust augmentation techniques that enhance model accuracy and resilience across diverse conditions, including fluctuating lighting and adverse weather. This novel adaptation ensures that the YOLO models can handle real-time detection demands with greater reliability and precision in complex, dynamic urban scenarios. Extensive experimentation demonstrates the effectiveness of this adaptation in improving mean average precision (mAP), recall, and F1 scores across various settings, with YOLOv10 emerging as the most versatile model, excelling in both precision and adaptability. The comparative analysis underscores this model’s suitability for urban traffic applications, where robust, adaptable object detection is critical. By addressing the unique challenges of traffic data, this study provides a foundational approach for developing more specialized, efficient deep learning models tailored to the specific needs of intelligent traffic monitoring systems