I am trying to implement this YOLO implementation for a purpose and it's giving wrong predictions. I am loading the same model as proposed by the author and following almost the same steps (as shown):
def _main(args):
yolo_model1 = load_model(model1) ###########LOADS YOLO
sess = K.get_session()
'''################ FUNCTION TO LOAD MY MODEL ############################'''
def get_model():
# load json and create model
json_file = open('/home/sarvagya/Desktop/RBC/Python/time-to-collision/trucks/model.json', 'r')
loaded_model_json = json_file.read()
json_file.close()
loaded_model = model_from_json(loaded_model_json)
# load weights into new model
config = tf.ConfigProto()
config.gpu_options.per_process_gpu_memory_fraction = 0.5
config.gpu_options.visible_device_list = "0"
tf.Session(config=config)
loaded_model.load_weights("/home/sarvagya/Desktop/RBC/Python/time-to-collision/trucks/model.h5")
loaded_model.compile(loss='binary_crossentropy',
optimizer='rmsprop',
metrics=['accuracy'])
return loaded_model
loaded_model = get_model() ############LOADS MY MODEL
anchors_path = os.path.expanduser(args.anchors_path)
classes_path = os.path.expanduser(args.classes_path)
print(classes_path)
with open(classes_path) as f:
class_names = f.readlines()
class_names = [c.strip() for c in class_names]
# print(class_names)
with open(anchors_path) as f:
anchors = f.readline()
anchors = [float(x) for x in anchors.split(',')]
anchors = np.array(anchors).reshape(-1, 2)
num_classes = len(class_names)
num_anchors = len(anchors)
yolo_outputs = yolo_head(yolo_model1.output, anchors, len(class_names))
# print("class names: ")
# print(yolo_outputs)
input_image_shape = K.placeholder(shape=(2, ))
boxes, scores, classes = yolo_eval(
yolo_outputs,
input_image_shape,
score_threshold=args.score_threshold,
iou_threshold=args.iou_threshold)
img = cv2.imread(input_img)
x1 = 416/img.shape[1]
y1 = 416/img.shape[0]
height, width = img.shape[0],img.shape[1]
new_image_size = (width - (width % 32),
height - (height % 32))
resized_image = cv2.resize(img, None, fx = x1, fy = y1, interpolation = cv2.INTER_AREA)
# print(type(resized_image))
resized_image = np.expand_dims(resized_image, 0)
out_boxes, out_scores, out_classes = sess.run(
[boxes, scores, classes],
feed_dict={
yolo_model1.input: resized_image,
input_image_shape: [width, height],
K.learning_phase(): 0
})
print('Found {} boxes for {}'.format(len(out_boxes), input_img))
print(out_classes)
for i, c in reversed(list(enumerate(out_classes))):
predicted_class = class_names[c]
box = out_boxes[i]
score = out_scores[i]
label = '{} {:.2f}'.format(predicted_class, score)
top, left, bottom, right = box
top = max(0, np.floor(top + 0.5).astype('int32'))
left = max(0, np.floor(left + 0.5).astype('int32'))
bottom = min(resized_image.shape[1], np.floor(bottom + 0.5).astype('int32'))
right = min(resized_image.shape[0], np.floor(right + 0.5).astype('int32'))
print(label, (left, top), (right, bottom))
Yet, I am getting wrong predictions. Can someone help me?
Hi,
I am trying to implement this YOLO implementation for a purpose and it's giving wrong predictions. I am loading the same model as proposed by the author and following almost the same steps (as shown):
Yet, I am getting wrong predictions. Can someone help me?