图像检测网络

$$图像检测网络$$

---

Realize:代码实现

---

Teach:代码讲解

深度学习—图像检测Demo代码（使用PyTorch进行训练、预测、TensorRT部署

---

一些检测常识

---

通用目标检测算法结构

1. Input：Image、Patches、Image Pyramid
2. Backbone:VGG、Resnet、EfficientNet
3. Neck:
   <1>Additional blocks:SPP、ASPP、RFB、SAM
   <2>Path-aggregation blocks:FPN、PAN、NAS-FPN、Fully-connected FPN、BiFPN、ASFF、SFAM
4. Heads
   <1>Dense Prediction(one-stage)
   <2>Sparse Prediction (two-stage)

---

常用的检测评价指标

---

2013-2018:

Loss:
1.RetinaNet-2017

OneStage:

YOLO系列

1.YOLO v1
2.YOLO v2-2016
3.YOLO v3-2018
4.YOLO-v4-2020
5.YOLO-v5（伪）-2020

SSD系列

6.SSD
7.DSSD

---

1.FCOS

其他

8.CenterNet-2019
9.SNIP-2018
9.M2Det-2019
10.DetectoRs-2020
11.EfficientDet-2020
12.CBNet-2020
13.TSD-2020
14.ResNest-2020

TwoStage:

Faster RCNN系列

1.RCNN
2.Fast RCNN
3.Faster RCNN
4.Mask R-CNN-2017
5.Cascade RCNN

---

Transformer

1. DETR
2. MaskFormer