1、可以做成 github+wiki ?
大家分享表情图 /斗图;喜欢的就 fork 到自己的库 ^_^
2、对于表情,可以用公有云的认知服务遍历一遍,既能拿到面部区域,也能拿到常见心情的置信度:
JObject rect = new JObject();
rect["left"] = er.FaceRectangle.Left;
rect["top"] =
er.FaceRectangle.Top;
rect["width"] = er.FaceRectangle.Width;
rect["height"] = er.FaceRectangle.Height;
JObject scores = new JObject();
scores["anger"] = er.Scores.Anger;
scores["contempt"] = er.Scores.Contempt;
scores["disgust"] = er.Scores.Disgust;
scores["fear"] = er.Scores.Fear;
scores["happiness"] = er.Scores.Happiness;
scores["neutral"] = er.Scores.Neutral;
scores["sadness"] = er.Scores.Sadness;
scores["surprise"] = er.Scores.Surprise;
JObject emotion = new JObject();
emotion["faceRectangle"] = rect;
emotion["scores"] = scores;
3、使用用公有云的 API 对每个表情图做个 OCR 处理