这是我的YOLOv8模型的改进项目,预计会有一百多个改进机制 目前有哪些 1.第一讲 增加一个小目标检测头: https://www.bilibili.com/video/BV1ErxWeWEAo/?vd_source=8a6043a22d94a87da35299c073140577 2.第二讲 增加非局部注意力机制NonLocalBlockND:https://www.bilibili.com/video/BV1Nm4KeJEG4/?vd_source=8a6043a22d94a87da35299c073140577 3.第三讲 增加多核感知网络PKINet:https://www.bilibili.com/video/BV1rc4Fe4Eh8/?spm_id_from=333.999.0.0&vd_source=8a6043a22d94a87da35299c073140577 4.第四讲 增加阶门控聚合模块Multi-Order Gated Aggregation和通道聚合模块CA block:https://www.bilibili.com/video/BV1ua2EYoEp4/?spm_id_from=333.999.0.0&vd_source=8a6043a22d94a87da35299c073140577 5.第五讲 增加StarNet星形卷积StarsBlock:https://www.bilibili.com/video/BV15P2MYKEWv/?spm_id_from=333.999.0.0&vd_source=8a6043a22d94a87da35299c073140577 6.第六讲 增加多尺度卷积注意力模块MSCAM:https://www.bilibili.com/video/BV13Y2uYKEhB/?spm_id_from=333.999.0.0&vd_source=8a6043a22d94a87da35299c073140577 7.第七讲 增加 一种新颖的注意力机制 Outlook Attention:https://www.bilibili.com/video/BV1iemgY7EWS/?spm_id_from=333.999.0.0&vd_source=8a6043a22d94a87da35299c073140577 8.第八讲 增加上下文引导模块 ContextGuided:https://www.bilibili.com/video/BV1oamgYmEMd/?spm_id_from=333.999.0.0&vd_source=8a6043a22d94a87da35299c073140577 9.第九讲 添加EMAttention注意力机制 https://www.bilibili.com/video/BV1ZtycYUE1i/?vd_source=8a6043a22d94a87da35299c073140577 10.第十讲 添加全维度动态卷积(Omni-dimensional Dynamic Convolution,ODConv)https://www.bilibili.com/video/BV1hyyAY1Ere/?vd_source=8a6043a22d94a87da35299c073140577 11.第十一讲 添加自适应阈值焦点损失(ATFL)函数解决类别不平衡 https://www.bilibili.com/video/BV1PfyWYoE7J/?vd_source=8a6043a22d94a87da35299c073140577 12.第十二讲 添加Haar小波下采样Down_wt卷积 https://www.bilibili.com/video/BV1sZyWYtEJR/?vd_source=8a6043a22d94a87da35299c073140577 13.第十三讲 添加卷积和注意力融合模块(CAFM) 提升小目标和遮挡检测 https://www.bilibili.com/video/BV1kcysYsEdJ/?vd_source=8a6043a22d94a87da35299c073140577