Skip to content

识别图片坐标误差很大,只是打开美团,淘宝app 这种 操作,实现不了 #246

@yangxinQA

Description

@yangxinQA

python run_mobileagentv3.py --adb_path "/Users/yangxin/Library/Android/sdk/platform-tools/adb" --api_key "sk-03f0c80673ec49949c62c8bc24b5410d" --base_url "https://dashscope.aliyuncs.com/compatible-mode/v1" --model "qwen3-vl-plus" --instruction "open app(淘宝)" --add_info "Some supplementary knowledge, can also be empty"

返回日志点击的点
Action: {"action": "click", "coordinate": [375, 241]}
Action: {"action": "click", "coordinate": [384, 237]}
Action: {"action": "click", "coordinate": [426, 910]}
Action: {"action": "long_press", "coordinate": [384, 237]}
Action: {"action": "swipe", "coordinate": [499, 738], "coordinate2": [499, 306]}
Action: {"action": "swipe", "coordinate": [499, 738], "coordinate2": [499, 290]}
Action: {"action": "click", "coordinate": [384, 237]}

实际测量位置:左上角位置(400,551)右下角(548,730)
不知道为什么识别不到。用OCR 能识别到么?有什么api调用么?

Metadata

Metadata

Assignees

No one assigned

    Labels

    No labels
    No labels

    Type

    No type

    Projects

    No projects

    Milestone

    No milestone

    Relationships

    None yet

    Development

    No branches or pull requests

    Issue actions