CCTV-Gun
收藏资源简介:
CCTV-Gun是由石溪大学精心构建和标注的基准数据集,专注于从闭路电视(CCTV)图像中检测手枪。该数据集包含7767张来自不同CCTV摄像头和场景的图像,主要关注手枪,这是枪支犯罪中最常用的类型。数据集的创建过程涉及从三个公开可用的数据集(Monash Gun Dataset, US Real-time Gun detection dataset, UCF Crime Scene dataset)中精选图像,并提供详细的边界框标注,包括人物、手枪及其持有者。此外,每张图像都标有挑战因素,如模糊和遮挡,以帮助分析检测算法的性能。CCTV-Gun数据集的应用领域主要集中在提高公共安全,通过自动快速检测CCTV图像中的手枪,预防枪支相关暴力事件的发生。
CCTV-Gun is a benchmark dataset carefully constructed and annotated by Stony Brook University, dedicated to pistol detection from closed-circuit television (CCTV) images. This dataset comprises 7767 images sourced from diverse CCTV cameras and scenarios, with a primary focus on pistols, the most commonly utilized type of firearm in gun-related criminal activities. The dataset was developed by curating images from three publicly available datasets: the Monash Gun Dataset, the US Real-time Gun Detection Dataset, and the UCF Crime Scene Dataset, and provides detailed bounding box annotations for individuals, pistols and their holders. Additionally, each image is labeled with challenging factors such as blurriness and occlusion to facilitate performance analysis of detection algorithms. The core application of the CCTV-Gun dataset is to improve public safety by automatically and rapidly detecting pistols in CCTV images to prevent gun-related violent incidents.

- 1CCTV-Gun: Benchmarking Handgun Detection in CCTV Images石溪大学 · 2023年



