deep-learning - 如何使用 SSD512 获得自定义 object 检测的预训练权重？

# Object Detection using SSD512(Single Shot Multibox Detector) # Libraries from keras import backend as K from keras.preprocessing import image from keras.optimizers import Adam import imageio import numpy as np import os import cv2 # Dependencies from models.keras_ssd512 import ssd_512 from keras_loss_function.keras_ssd_loss import SSDLoss # Defining the width and height of the image height = 512 width = 512 # Defining confidence threshold confidence_threshold = 0.5 # Different Classes of objects in VOC dataset classes = ['background', 'aeroplane', 'bicycle', 'bird', 'boat', 'bottle', 'bus', 'car', 'cat', 'chair', 'cow', 'diningtable', 'dog', 'horse', 'motorbike', 'person', 'pottedplant', 'sheep', 'sofa', 'train', 'tvmonitor'] K.clear_session() # Clear previous models from memory. # creating model and loading pretrained weights model = ssd_512(image_size=(height, width, 3), # dimensions of the input images (fixed for SSD512) n_classes=20, # Number of classes in VOC 2007 & 2012 dataset mode='inference', l2_regularization=0.0005, scales=[0.07, 0.15, 0.3, 0.45, 0.6, 0.75, 0.9, 1.05], aspect_ratios_per_layer=[[1.0, 2.0, 0.5], [1.0, 2.0, 0.5, 3.0, 1.0/3.0], [1.0, 2.0, 0.5, 3.0, 1.0/3.0], [1.0, 2.0, 0.5, 3.0, 1.0/3.0], [1.0, 2.0, 0.5, 3.0, 1.0/3.0], [1.0, 2.0, 0.5], [1.0, 2.0, 0.5]], two_boxes_for_ar1=True, steps=[8, 16, 32, 64, 128, 256, 512], offsets=[0.5, 0.5, 0.5, 0.5, 0.5, 0.5, 0.5], clip_boxes=False, variances=[0.1, 0.1, 0.2, 0.2], normalize_coords=True, subtract_mean=[123, 117, 104], swap_channels=[2, 1, 0], confidence_thresh=0.5, iou_threshold=0.45, top_k=200, nms_max_output_size=400) # path of the pre trained model weights weights_path = 'weights\\VGG_VOC0712Plus_SSD_512x512_ft_iter_160000.h5' model.load_weights(weights_path, by_name=True) # Compiling the model adam = Adam(lr=0.001, beta_1=0.9, beta_2=0.999, epsilon=1e-08, decay=0.0) ssd_loss = SSDLoss(neg_pos_ratio=3, alpha=1.0) model.compile(optimizer=adam, loss=ssd_loss.compute_loss) # Paths of input and output inages input_image_path = 'inputs/images' output_image_path = 'outputs/images' # Paths of input and output videos input_video_path = 'inputs/videos' output_video_path = 'outputs/videos' # Transforming image size def transform(input_image): return cv2.resize(input_image, (512, 512), interpolation = cv2.INTER_CUBIC) # Function to detect objects in image def detect_object(original_image): original_image_height, original_image_width = original_image.shape[:2] input_image = transform(original_image) input_image = np.reshape(input_image, (1, 512, 512, 3)) y_pred = model.predict(input_image) actual_prediction = [y_pred[k][y_pred[k,:,1] > confidence_threshold] for k in range(y_pred.shape[0])] for box in actual_prediction[0]: # Coordinates of diagonal points of bounding box x0 = box[-4] * original_image_width / width y0 = box[-3] * original_image_height / height x1 = box[-2] * original_image_width / width y1 = box[-1] * original_image_height / height label_text = '{}: {:.2f}'.format(classes[int(box[0])], box[1]) # label text cv2.rectangle(original_image, (int(x0), int(y0)), (int(x1), int(y1)), (255, 0, 0), 2) # drwaing rectangle cv2.putText(original_image, label_text, (int(x0), int(y0)), cv2.FONT_HERSHEY_DUPLEX, 1, (231, 237, 243), 2, cv2.LINE_AA) # putting lable return original_image # Detecting objects in images for file in os.listdir(input_image_path): print('Reading', file) original_image = imageio.imread(os.path.join(input_image_path, file)) # Reading image if original_image is not None: output_image = detect_object(original_image) # detecting objects imageio.imwrite(os.path.join(output_image_path, file), output_image[:, :, :]) # savinng back images """