Chaquopy集成OpenCV无法通过cv2.VideoCapture打开URL链接

本地运行正常的目标识别Python代码，通过Chaquopy集成到Android应用后无法运行，故障触发点是Python代码中打开摄像头流的语句 cap = cv2.VideoCapture(url) 在Android设备上执行失效。

MainActivity.java #

import androidx.appcompat.app.AppCompatActivity;

import android.os.Bundle;
import android.widget.TextView;

import com.chaquo.python.PyObject;
import com.chaquo.python.Python;
import com.chaquo.python.android.AndroidPlatform;

public class MainActivity extends AppCompatActivity {

    @Override
    protected void onCreate(Bundle savedInstanceState) {
        super.onCreate(savedInstanceState);
        setContentView(R.layout.activity_main);
        TextView textView = findViewById(R.id.textview);

        if(!Python.isStarted())
            Python.start(new AndroidPlatform(this));

        Python py = Python.getInstance();
        PyObject pyobj = py.getModule("main");
        textView.setText(pyobj.callAttr("main").toString());
    }
}

main.py #

import cv2
import numpy as np

def main():
    thres = 0.50
    nms_threshold = 0.20
    cap = cv2.VideoCapture('http://192.168.137.10:8081/')
    classNames= []
    classFile = 'coco.names'
    with open(classFile,'rt') as f:
        classNames=[line.rstrip() for line in f]

    configPath = 'ssd_mobilenet_v3_large_coco_2020_01_14.pbtxt'
    weightsPath = 'frozen_inference_graph.pb'

    net = cv2.dnn_DetectionModel(weightsPath,configPath)
    net.setInputSize(300,300)
    net.setInputScale(1.0/ 127.5)
    net.setInputMean((127.5, 127.5, 127.5))
    net.setInputSwapRB(True)
    success,img = cap.read()
    classIds, confs, bbox = net.detect(img,confThreshold=thres)
    bbox = list(bbox)
    confs = list(np.array(confs).reshape(1, -1)[0])
    confs = list(map(float, confs))

    indices = cv2.dnn.NMSBoxes(bbox,confs,thres,nms_threshold)

    for i in indices:
        i=i[0]
        return (classNames[classIds[i][0] - 1])

2022-06-07 22:19:43.083 26043-26043/com.example.thirdeye E/cv::error(): OpenCV(4.5.1) Error: Bad argument (CAP_IMAGES: can't find starting number (in the name of file): http://192.168.137.10:8081/) in icvExtractPattern, file /home/smith/git/chaquo/python/server/pypi/packages/opencv-contrib-python/build/4.5.1.48/cp38-cp38-android_16_armeabi_v7a/src/opencv/modules/videoio/src/cap_images.cpp, line 253
2022-06-07 22:19:43.088 26043-26043/com.example.thirdeye E/OpenCV/4.5.1: [ERROR:0] global /home/smith/git/chaquo/python/server/pypi/packages/opencv-contrib-python/build/4.5.1.48/cp38-cp38-android_16_armeabi_v7a/src/opencv/modules/videoio/src/cap.cpp (142) open VIDEOIO(CV_IMAGES): raised OpenCV exception:
    
    OpenCV(4.5.1) /home/smith/git/chaquo/python/server/pypi/packages/opencv-contrib-python/build/4.5.1.48/cp38-cp38-android_16_armeabi_v7a/src/opencv/modules/videoio/src/cap_images.cpp:253: error: (-5:Bad argument) CAP_IMAGES: can't find starting number (in the name of file): http://192.168.137.10:8081/ in function 'icvExtractPattern'
2022-06-07 22:19:43.089 26043-26043/com.example.thirdeye A/libc: Fatal signal 7 (SIGBUS), code 1 (BUS_ADRALN), fault addr 0xdc40d601 in tid 26043 (xample.thirdeye), pid 26043 (xample.thirdeye)

从报错日志可以直接定位根因：Chaquopy源内置的OpenCV预编译包没有集成FFmpeg依赖，不支持HTTP、RTSP这类网络视频流的解析读取。OpenCV找不到匹配网络流的处理后端时，会自动回退到本地图片序列读取模式（CAP_IMAGES），尝试把传入的URL当成本地带数字序号的图片文件名解析，自然抛出“找不到文件名起始数字”的参数错误，后续空帧传入检测逻辑还触发了内存对齐异常，直接导致应用崩溃。

• 优先选实现成本最低的方案：不在Python层直接调用OpenCV拉取网络流，改用Android原生的网络请求/摄像头API获取视频帧，将帧数据通过Chaquopy的接口传给Python层，只保留Python里的目标检测推理逻辑，绕开OpenCV在Android端的视频流读取能力缺陷。
• 如果必须在Python层直接读流：自行交叉编译开启FFmpeg支持的对应CPU架构OpenCV Android二进制包，替换Chaquopy默认安装的无FFmpeg版本的opencv-python包，注意需要匹配Chaquopy使用的Python版本和目标设备的ABI架构。
• 补全容错逻辑：现有代码在cap.read()后没有判断success返回值，一旦帧读取失败，img为None时传入检测接口必然触发崩溃，无论选哪种方案都要加状态判断，读取失败直接返回错误，不要继续执行后续检测逻辑。

内容的提问来源于stack exchange，提问作者Sakthi Vinnavan