配套课件 : https://gf.bilibili.com/item/detail/1103548011
-
tensor相关
- tenso r内部存储结构 :https://www.bilibili.com/video/BV1NH4y177VL/
- tensor 的连续性 与 contiguous()方法 : https://www.bilibili.com/video/BV1Fz421R7T9/
-
模型可视化
- torchsummary : https://www.bilibili.com/video/BV1ZH4y1f7T1/
- netron : https://www.bilibili.com/video/BV1rh411M7JP/
- tensorboardX : https://www.bilibili.com/video/BV1BV4y1C7HJ/
-
模型搭建
- 转置卷积:https://www.bilibili.com/video/BV1M84y1T7tY/
- 膨胀卷积:https://www.bilibili.com/video/BV1za4y1K7E6/
- 可变形卷积 :https://www.bilibili.com/video/BV1Sh4y1y75i/
- Batch Normlization:https://www.bilibili.com/video/BV11s4y1c7pg/
- BatchNorm、LayerNorm、GroupNorm:https://www.bilibili.com/video/BV1UG411f7DL/
- nn.Sequential、nn.ModuleList、nn.ModuleDict:https://www.bilibili.com/video/BV1zu4y1i7Bk/
-
优化器(SGD、Momentum、Adagrad、RMSProp、Adam):https://www.bilibili.com/video/BV1jh4y1q7ua/
-
损失函数
- 交叉熵损失函数 : https://www.bilibili.com/video/BV1Sm4y127kY/
- 信息量、熵、交叉熵、KL散度、交叉熵损失函数 : https://www.bilibili.com/video/BV1L8411X7ZZ/
- 目标检测 - 定位损失 : https://www.bilibili.com/video/BV1Bw411J7R6/
- focal loss💰:https://www.bilibili.com/video/BV1jG411B7X9/
-
插值算法
-
数据预处理&数据增强
-
Dataset、Dataloade - 上 :https://www.bilibili.com/video/BV1vV4y1v7LY/
-
Dataset、Dataloade - 下:https://www.bilibili.com/video/BV1bM411L7Qi/
-
图像预处理/图像增强 - 上:https://www.bilibili.com/video/BV1Uw411A7eK/
-
图像预处理/图像增强 - 下:https://www.bilibili.com/video/BV1ew411k7UP/
-
-
数据集
- MSCOCO数据集:https://www.bilibili.com/video/BV1GT411W7ji/
- Pascal VOC:https://www.bilibili.com/video/BV18s4y1v7FN/
-
日志记录
- logging
- 1、基础使用、https://www.bilibili.com/video/BV1e34y1M7Ps/
- 2、高级应用 (购买课件后查看)
- 3、配置文件 (购买课件后查看)
- tensorboard / tensorboardX:
- 1、记录训练数据指标:https://www.bilibili.com/video/BV18w411P75R/
- 2、模型可视化、图像可视化 (购买课件后查看)
- wandb
- 1、安装与登录 (购买课件后查看)
- 2、基础使用与可视化 (购买课件后查看)
- 3、wandb.watch() 模型、参数、梯度的监控 (购买课件后查看)
- 4、wandb.Artifact() 版本管理、文件保存与下载 (购买课件后查看)
- logging
-
其他
- 迁移学习:https://www.bilibili.com/video/BV1X8411f7q1/
- IoU 与 NMS:https://www.bilibili.com/video/BV19L41187s4/
- 模型结构图 绘画工具 - Axure:https://www.bilibili.com/video/BV12u411c7PZ/
- Kaggle项目部署:https://www.bilibili.com/video/BV1qM41137SY/
💰 为充电专属视频
- self-attention:https://www.bilibili.com/video/BV1qo4y1F7Ep/
- Multi-head attention:https://www.bilibili.com/video/BV1gV4y167rS/
- MultiheadAttention 中的参数 key_padding_mask 与 attention_mask : https://www.bilibili.com/video/BV1Xg4y127fd/
- ViT:https://www.bilibili.com/video/BV1xm4y1b7Pw/
- Swin-transformer:https://www.bilibili.com/video/BV1Nz4y177aW/
-
Deformable DETR
- 1、Abstract 算法概述:https://www.bilibili.com/video/BV1rh4y1F7fG/
- 2、backbone 、MultiHeadAttention 公式讲解:https://www.bilibili.com/video/BV1vr4y1o73B/
- 3、Deformable Attention、MSDeformAttention、流程讲解:https://www.bilibili.com/video/BV1Uj411z7QU/
-
RT-DETR
- 1、abstract 算法概述:https://www.bilibili.com/video/BV1Nb4y1F7k9/
- 2、backbone:https://www.bilibili.com/video/BV1Ac411k7Sg/
- 3、AIFI:https://www.bilibili.com/video/BV1be411S74w/
- 4、CCFM:https://www.bilibili.com/video/BV1DG411z71m/
- 5、CCFM 收尾工作 💰:https://www.bilibili.com/video/BV1q64y1P7hR/
- 6、decoder 整体网络结构 💰:https://www.bilibili.com/video/BV1LK4y1z7op/
- 7、denoising 理论讲解 💰:https://www.bilibili.com/video/BV1AW4y1w79p/
- 8、denoising 代码讲解 💰:https://www.bilibili.com/video/BV1eK411e7Jh/
- 9、IoU-aware query selection 理论讲解 💰 :https://www.bilibili.com/video/BV1d64y1N7sg/
- 10、IoU-aware query selection 代码讲解 💰 :https://www.bilibili.com/video/BV1Z94y1K744/
- 11、decoder + MSDeformableAttention 💰:https://www.bilibili.com/video/BV1mZ4y1J7xA/
- 12、损失函数 💰 : https://www.bilibili.com/video/BV1uv421i7gC/
-
相关知识
-
玩转resnet50:https://www.bilibili.com/video/BV1Ya4y1o7HP/
-
Attention Map 💰:https://www.bilibili.com/video/BV1jC411W7sC/
-
Positional Embedding 💰:https://www.bilibili.com/video/BV1z84y1R7t8/
-
-
yolo v1- 上:https://www.bilibili.com/video/BV1ec411T7Rh/
-
yolo v1- 下:https://www.bilibili.com/video/BV1ks4y137oU/
-
DDPM 理论知识:
- 1 - 概述:https://www.bilibili.com/video/BV1Xu4y147ov/
- 2 - 基础知识准备:https://www.bilibili.com/video/BV1UN411E7hz/
- 3 - 前向扩散过程:https://www.bilibili.com/video/BV1vw41117r1/
- 4 - 反向去噪过程、图像概率分布的理解:https://www.bilibili.com/video/BV13N411J7zK/
- 5 - 损失函数:https://www.bilibili.com/video/BV1ju4y1x7L4/
- 6 - 代码精讲 💰:https://www.bilibili.com/video/BV1xc411X7wx/