fix: VLM prompt 改为通用食物识别,不限于中餐
This commit is contained in:
@@ -270,11 +270,10 @@ public static class AiChatEndpoints
|
||||
}
|
||||
|
||||
var prompt = """
|
||||
识别图片中所有食物,只返回JSON,不要加任何其他文字:
|
||||
识别图片中所有食物,用中文名称,只返回JSON:
|
||||
{
|
||||
"foods": [{"name":"中文菜名","portion":"份量","calories":整数}]
|
||||
"foods": [{"name":"食物名","portion":"份量","calories":整数}]
|
||||
}
|
||||
常见参考:米饭200g/230卡,炒青菜200g/80卡,红烧肉150g/400卡,番茄炒蛋200g/180卡,青椒肉丝200g/200卡。
|
||||
""";
|
||||
|
||||
try
|
||||
|
||||
Reference in New Issue
Block a user