无法通过视频帧进行对象检测 Tensorflow python答案

【问题标题】：Unable to Pass Video Frame for Object Detection Tensorflow python无法通过视频帧进行对象检测 Tensorflow python
【发布时间】：2020-06-05 01:43:03
【问题描述】：

我可以通过将帧作为图像然后处理来处理视频帧。但无法直接传帧到物体检测。使用 imwrite 保存图像会使程序变慢...

这是我的主要方法：

cap = cv2.VideoCapture(gstreamer_pipeline(flip_method=2), cv2.CAP_GSTREAMER)

if cap.isOpened():
    window_handle = cv2.namedWindow("CSI Camera", cv2.WINDOW_AUTOSIZE)
    # Window
    while cv2.getWindowProperty("CSI Camera", 0) >= 0:
        ret_val, frame = cap.read()
        if not ret_val:
            break
        frame = imutils.resize(frame, width=600)

        #cv2.imwrite('box.jpg', frame)
        #image = Image.open(path)
        #Error in here!!!
        predictions = od_model.predict_image(frame)


        for x in range(len(predictions)):
            probab = (predictions[x]['probability'])*100
            if(probab > 45):
                print(predictions[x]['tagName'], end=' ')
                print(probab)
        #cv2.imshow("CSI Camera", frame)
        # This also acts as
        keyCode = cv2.waitKey(30) & 0xFF
        # Stop the program on the ESC key
        if keyCode == 27:
            break
    cap.release()
    cv2.destroyAllWindows()
else:
    print("Unable to open camera")

错误信息：

predictions = od_model.predict_image(frame)
File "/home/bharat/New_IT3/object_detection.py", line 125, in 
predict_image
inputs = self.preprocess(image)
File "/home/bharat/New_IT3/object_detection.py", line 130, in 
preprocess
image = image.convert("RGB") if image.mode != "RGB" else image
AttributeError: 'numpy.ndarray' object has no attribute 'mode'

【问题讨论】：

如果你取消注释#cv2.imwrite('box.jpg', frame) #image = Image.open(path) then??
然后它正在工作，但一次又一次地将图像写入内存会减慢进程。我正在寻找直接传递对象
from PIL import Image im = Image.fromarray(frame) //导入PIL包

标签： opencv tensorflow object-detection nvidia-jetson-nano

【解决方案1】：

open cv 读取 bgr 色谱中的图像将其转换为 rgb 并发送图像进行检测，api 相同 -

frame = cv2.cvtColor(frame, cv2.COLOR_BGR2RGB)

【讨论】：