This paper introduces CNTCB-YOLOv7, an effective forest fire detection model based on ConvNeXtV2 and CBAM. The study aims to enhance the feature representation and detection accuracy of the YOLOv7 algorithm in large-scale fire areas and complex forest environments. ConvNeXtV2 and Conv2Former are integrated into YOLOv7 to improve performance. The backbone network is enhanced with four attention mechanisms: NAM, SimAM, GAM, and CBAM, leading to the development of the CNTCB-YOLOv7 algorithm. Experimental results show that CNTCB-YOLOv7 outperforms YOLOv7 with improved accuracy (2.39%), recall rate (0.73%), and average precision (AP) (1.14%). The model also has a reduced parameter count, improving inference speed. The CNTCB-YOLOv7 algorithm demonstrates better detection performance in large-scale and complex forest fire scenarios, with enhanced focus on critical information. The model's performance is validated through experiments on a dataset of 2590 images, showing improved detection accuracy and reduced false positives. The study highlights the effectiveness of combining ConvNeXtV2 and CBAM in forest fire detection, contributing to more efficient and accurate fire monitoring and management. Future work includes further refinement of the model, comparison with other detection algorithms, and expansion to diverse datasets and applications.This paper introduces CNTCB-YOLOv7, an effective forest fire detection model based on ConvNeXtV2 and CBAM. The study aims to enhance the feature representation and detection accuracy of the YOLOv7 algorithm in large-scale fire areas and complex forest environments. ConvNeXtV2 and Conv2Former are integrated into YOLOv7 to improve performance. The backbone network is enhanced with four attention mechanisms: NAM, SimAM, GAM, and CBAM, leading to the development of the CNTCB-YOLOv7 algorithm. Experimental results show that CNTCB-YOLOv7 outperforms YOLOv7 with improved accuracy (2.39%), recall rate (0.73%), and average precision (AP) (1.14%). The model also has a reduced parameter count, improving inference speed. The CNTCB-YOLOv7 algorithm demonstrates better detection performance in large-scale and complex forest fire scenarios, with enhanced focus on critical information. The model's performance is validated through experiments on a dataset of 2590 images, showing improved detection accuracy and reduced false positives. The study highlights the effectiveness of combining ConvNeXtV2 and CBAM in forest fire detection, contributing to more efficient and accurate fire monitoring and management. Future work includes further refinement of the model, comparison with other detection algorithms, and expansion to diverse datasets and applications.