Kivy Android应用实时帧显示黑屏问题及色彩格式解决方案

# ... (接收并反序列化帧数据)
frame = pickle.loads(frame_data)
buffer = cv2.flip(frame, 0).tobytes()
texture = Texture.create(size=(frame.shape[1], frame.shape[0]), colorfmt='bgr') # 问题所在
texture.blit_buffer(buffer, colorfmt='bgr', bufferfmt='ubyte') # 问题所在
self.image.texture = texture

# ... (接收并反序列化帧数据)
frame = pickle.loads(frame_data)
# 注意：OpenCV的frame默认是BGR。如果Kivy在Android上期望RGB，
# 且仅通过colorfmt='rgb'声明就能解决黑屏，
# 那么Kivy可能在内部处理了BGR到RGB的转换，或者'bgr'声明本身在Android上不被支持。
# 如果后续出现颜色反转，则需要在此处添加 cv2.cvtColor(frame, cv2.COLOR_BGR2RGB)
buffer = cv2.flip(frame, 0).tobytes()
texture = Texture.create(size=(frame.shape[1], frame.shape[0]), colorfmt='rgb') # 修改为 'rgb'
texture.blit_buffer(buffer, colorfmt='rgb', bufferfmt='ubyte') # 修改为 'rgb'
self.image.texture = texture

from kivymd.app import MDApp
from kivy.uix.image import Image
from kivy.clock import Clock
from kivy.graphics.texture import Texture
import socket
import cv2
import pickle
import struct
# ... 其他导入

class Angelus(MDApp):
    # ... build, show_popup, on_ok 等方法保持不变

    def update_frame(self, dt):
        # ... (数据接收逻辑保持不变)
        while len(self.data) < self.payload_size:
            packet = self.client_socket.recv(4 * 1024)
            if not packet: break
            self.data += packet
        packet_msg_size = self.data[:self.payload_size]
        self.data = self.data[self.payload_size:]
        msg_size = struct.unpack("Q", packet_msg_size)[0]

        while len(self.data) < msg_size:
            self.data += self.client_socket.recv(4 * 1024)
        frame_data = self.data[:msg_size]
        self.data = self.data[msg_size:]

        frame = pickle.loads(frame_data)

        # 核心修正：将色彩格式声明从 'bgr' 改为 'rgb'
        buffer = cv2.flip(frame, 0).tobytes()
        texture = Texture.create(size=(frame.shape[1], frame.shape[0]), colorfmt='rgb')
        texture.blit_buffer(buffer, colorfmt='rgb', bufferfmt='ubyte')
        self.image.texture = texture

    # ... update_data 方法保持不变

Angelus().run()

import cv2
import numpy as np
import pickle
import struct
import socket
import threading
# ... 其他导入和TensorFlow/对象检测相关代码

def send_frames(image_np_with_detections, client_socket):
    a = pickle.dumps(image_np_with_detections)
    message = struct.pack("Q", len(a)) + a
    client_socket.sendall(message)

# ... (服务器初始化和模型加载)

while cap.isOpened():
    ret, frame = cap.read()
    image_np = np.array(frame)
    if image_np is not None:
        # ... (对象检测和可视化处理)
        # image_np_with_detections 此时是OpenCV格式的图像（通常为BGR）
        client_thread = threading.Thread(target=send_frames, args=(image_np_with_detections, client_socket))
        client_thread.start()
        # ... (其他数据发送和退出逻辑)