法蘭斯 Fustyles AI 專區
[物件辨識 (coco-ssd)] (Single Shot Multibox Detector)
註1：需用Chrome開啟，與攝影鏡頭，微軟COCO數據集包括90類目標(列表)，手機可以切換鏡頭
註2：何謂SSD說明
圖片 https://fustyles.github.io/webduino/TensorFlow/ObjectDetection_image/ObjectDetection_image_coco-ssd.html
視訊 (Chrome瀏覽器) https://fustyles.github.io/webduino/TensorFlow/ObjectDetection_video/ObjectDetection_video_coco-ssd.html[物件辨識 (mobilenet)]
註：兩個偵測的物件種類不同，COCO可取得物件位置大小(框選)，MobileNet則否，但辨識種類多(列表)
圖片 https://fustyles.github.io/webduino/TensorFlow/ObjectDetection_image/ObjectDetection_image_mobilenet.html
視訊(Chrome瀏覽器) https://fustyles.github.io/webduino/TensorFlow/ObjectDetection_video/ObjectDetection_video_mobilenet.html[姿態辨識 posenet]
圖片(單人) https://fustyles.github.io/webduino/TensorFlow/PoseDetection_image/PoseDetection_image.html
圖片(多人) https://fustyles.github.io/webduino/TensorFlow/PoseDetection_image/PoseDetection_image_multi.html
視訊(單人) https://fustyles.github.io/webduino/TensorFlow/PoseDetection_video/PoseDetection_video.html
視訊(多人) https://fustyles.github.io/webduino/TensorFlow/PoseDetection_video/PoseDetection_video_multi.html[BodyPix V1]
圖片 https://fustyles.github.io/webduino/TensorFlow/BodyPix_image/BodyPix_image.html
視訊(Chrome瀏覽器) https://fustyles.github.io/webduino/TensorFlow/BodyPix_video/BodyPix_video.html[BodyPix V2]
圖片 https://fustyles.github.io/webduino/TensorFlow/BodyPix_image/BodyPix2_image.html
視訊(Chrome瀏覽器) https://fustyles.github.io/webduino/TensorFlow/BodyPix_video/BodyPix2_video.html[DeepLab v3]
圖片 https://fustyles.github.io/webduino/TensorFlow/ObjectDetection_image/ObjectDetection_image_deeplab.html
視訊(Chrome瀏覽器) https://fustyles.github.io/webduino/TensorFlow/ObjectDetection_video/ObjectDetection_video_deeplab.html[手寫或匯入圖片訓練辨識 knn-classifier] 基於KNN分類演算法說明
https://fustyles.github.io/webduino/TensorFlow/DigitRecognition_knn-classifier/DigitRecognition_knn-classifier.html[視訊深度學習 knn-classifier] 自訂分類辨識(最多可建立10種模型，可以存檔)
https://fustyles.github.io/webduino/TensorFlow/VideoRecognition_knn-classifier/VideoRecognition_knn-classifier.html[Face Detection (face-api.js)
圖片 https://fustyles.github.io/webduino/TensorFlow/Face-api/Face-api_FaceDetection_image.html
視訊(Chrome瀏覽器) https://fustyles.github.io/webduino/TensorFlow/Face-api/Face-api_FaceDetection_video.html[Face Recognition (face-api.js)]
圖片 https://fustyles.github.io/webduino/TensorFlow/Face-api/Face-api_FaceRecognition_image.html
視訊(Chrome瀏覽器) https://fustyles.github.io/webduino/TensorFlow/Face-api/Face-api_FaceRecognition_video.html[手勢辨識 Handpose]
圖片 https://fustyles.github.io/webduino/TensorFlow/HandPoseDetection_image/handpose_image.html
視訊(Chrome瀏覽器) https://fustyles.github.io/webduino/TensorFlow/HandPoseDetection_video/handpose_video.html[臉部網格 Facemesh]
圖片 https://fustyles.github.io/webduino/TensorFlow/FacemeshDetection_image/facemesh_image.html
視訊(Chrome瀏覽器) https://fustyles.github.io/webduino/TensorFlow/FacemeshDetection_video/facemesh_video.html[標記臉部 Blazeface]
圖片 https://fustyles.github.io/webduino/TensorFlow/Blazeface_image/Blazeface_image.html
視訊(Chrome瀏覽器) https://fustyles.github.io/webduino/TensorFlow/Blazeface_video/Blazeface_video.html[文字辨識]
圖片 https://fustyles.github.io/webduino/tesseract.js/Image2Text_image.html?back
視訊 https://fustyles.github.io/webduino/tesseract.js/Image2Text_video.html?back[顏色偵測] https://fustyles.github.io/webduino/Tracking.js/tracking.js_DetectColor_video.html
註：可自訂三種RGB範圍，當符合範圍的像素點眾多且聚集，即可被框選出來。[偵測臉部] https://fustyles.github.io/webduino/Tracking.js/tracking.js_DetectFace_video.html
可選擇不馬賽克對象：第一個被偵測到的人臉、人臉面積最大(最靠近鏡頭)的或全部馬賽克(需用Chrome開啟)
tfjs posenet 影像辨識功能的應用(需用Chrome開啟) 註：Google TensorFlow簡介
點選按鈕"Capture background image"捕捉無人時畫面→兩手腕高於鼻子隱形→兩手拜佛動作現形，需在肩膀下，人離鏡頭遠一點，因為太近看不到肩膀底下手腕位置