This paper presents YOLO-based models for detecting and localizing smoke and wildfires in ground and aerial images. The study evaluates recent YOLO models, including YOLOv7x, YOLOv8s, and others, for their performance in detecting smoke and wildfires. YOLOv7x achieved the highest mAP (mean Average Precision) score of 80.40% and fast detection speed, outperforming baseline models. YOLOv8s achieved a high mAP of 98.10% in identifying and localizing wildfire smoke. These models demonstrated significant potential in handling challenging scenarios, including detecting small fire and smoke areas, varying fire and smoke features, complex backgrounds, and visual similarities among smoke, fog, clouds, fire, lighting, and sun glare. The study uses two public datasets, D-Fire and WSDY, for evaluation. The D-Fire dataset includes 21,527 images of smoke, fire, and fire/smoke-like objects, while the WSDY dataset contains 737 smoke images. The evaluation metrics include mAP, GFLOPS (billions of floating-point operations per second), and inference time. The results show that YOLOv7x and YOLOv8s outperformed other models in terms of detection accuracy and speed. The lightweight models, such as YOLOv5n, YOLOv5s, YOLOv8n, and YOLOv8s, achieved high performance with faster inference times and lower computational costs. The study highlights the effectiveness of YOLO models in detecting and localizing smoke and wildfires in both ground and aerial images. These models can address challenges such as background complexity, detection of small fire and smoke areas, varying fire and smoke features, and visual similarities among smoke, fog, clouds, fire, lighting, and sun glare. The results demonstrate that YOLO models are suitable for real-time fire detection and monitoring, enabling rapid response and reducing environmental damage. Future work includes optimizing YOLO models for deployment on edge devices to improve wildland fire detection performance.This paper presents YOLO-based models for detecting and localizing smoke and wildfires in ground and aerial images. The study evaluates recent YOLO models, including YOLOv7x, YOLOv8s, and others, for their performance in detecting smoke and wildfires. YOLOv7x achieved the highest mAP (mean Average Precision) score of 80.40% and fast detection speed, outperforming baseline models. YOLOv8s achieved a high mAP of 98.10% in identifying and localizing wildfire smoke. These models demonstrated significant potential in handling challenging scenarios, including detecting small fire and smoke areas, varying fire and smoke features, complex backgrounds, and visual similarities among smoke, fog, clouds, fire, lighting, and sun glare. The study uses two public datasets, D-Fire and WSDY, for evaluation. The D-Fire dataset includes 21,527 images of smoke, fire, and fire/smoke-like objects, while the WSDY dataset contains 737 smoke images. The evaluation metrics include mAP, GFLOPS (billions of floating-point operations per second), and inference time. The results show that YOLOv7x and YOLOv8s outperformed other models in terms of detection accuracy and speed. The lightweight models, such as YOLOv5n, YOLOv5s, YOLOv8n, and YOLOv8s, achieved high performance with faster inference times and lower computational costs. The study highlights the effectiveness of YOLO models in detecting and localizing smoke and wildfires in both ground and aerial images. These models can address challenges such as background complexity, detection of small fire and smoke areas, varying fire and smoke features, and visual similarities among smoke, fog, clouds, fire, lighting, and sun glare. The results demonstrate that YOLO models are suitable for real-time fire detection and monitoring, enabling rapid response and reducing environmental damage. Future work includes optimizing YOLO models for deployment on edge devices to improve wildland fire detection performance.