CV學習筆記

Iris
Sep 17, 2021

--

1 影像分類(image classification): inceptionV3 + CAM /分類物件種類
2 物件偵測(Object localization): YOLOv4 /框出物件位置
3 語意分割(Semantic segmentation): U-Net /可割出物件輪廓
4 物件分割(instance segmentation)(2+3): Mask R-CNN
[補充] : region-based segmentation
*如果要將圖像分為兩個區域(對象和背景),則定義一個threshlod為global threshlod
*如果有多個對象以及背景,則必須定義多個threshlod為local threshlod

特殊範例
* 人臉辨識: FaceNet
* 姿態辨識: Human Pose
* 車牌辨識: LPRNet

--

--

Iris

Machine/Deep Learning, Image Processing and Computer Vision