1

在我的旧项目中，我使用了一个 yolov3 模型并对其进行了训练，并使用了 cv2.readNetFromDarknet(...)、cv2.dnn.blobFromImage(...) 和 net.forward(...) 来运行我的推论，效果很好。我训练了我的 yolov4 模型，并假设我只需要更改项目中的配置和权重文件即可进行推理，但我没有检测到边界框。除了权重和配置之外应该有变化吗？谢谢你。

编辑：

net = cv2.dnn.readNetFromDarknet(PATHS["model_config"], PATHS["model_weights"])
layer_names = net.getLayerNames()
output_layer_names = [ layer_names[index[0] - 1] for index in net.getUnconnectedOutLayers() ]

#during inference
def method(frame):
   blob = cv2.dnn.blobFromImage(frame, 1 / 255.0, (MAGIC_NUMBER_1, 
   MAGIC_NUMBER_1), swapRB=True, crop=False)
   net.setInput(blob)
   layerOutputs = net.forward(output_layer_names)

   boxes = []
   confidences = []
   classIDs = []

   for output in layerOutputs:
          for detection in output:
               scores = detection[5:]
               classID = np.argmax(scores)
               confidence = scores[classID]
            
               if confidence > .5: # NEVER PASSES, CONFIDENCE IS ALWAYS 0 WHEN PRINTED OUT
                
                   box = detection[0:4] * np.array([W, H, W, H])
                   (centerX, centerY, width, height) = box.astype("int")

                   x = int(centerX - (width / 2))
                   y = int(centerY - (height / 2))

                   boxes.append([x, y, int(width), int(height)])
    
                   confidences.append(float(confidence))
                   classIDs.append(classID)

   return boxes, confidences, classIDs
4

0 回答 0