what.models.detection.yolo.yolov3_tiny 

 1import cv2
 2import numpy as np
 3from keras.models import load_model
 4
 5from what.models.detection.utils.time_utils import Timer
 6
 7from .utils.yolo_utils import yolo_process_output, yolov3_tiny_anchors
 8
 9class YOLOV3_TINY:
10    def __init__(self, class_names, model_path):
11        self.model = load_model(model_path)
12        self.class_names = class_names
13        self.timer = Timer()
14
15    def predict(self, image, top_k=-1, prob_threshold=None):
16        input_cv_image = cv2.resize(image, (416, 416))
17        input_cv_image = np.array(input_cv_image).astype(np.float32) / 255.0
18
19        # Yolo inference
20        self.timer.start()
21        outs = self.model.predict(np.array([input_cv_image]))
22        print("FPS: ", int(1.0 / self.timer.end()))
23
24        boxes, class_ids, confidences = yolo_process_output(outs, yolov3_tiny_anchors, len(self.class_names))
25
26        return input_cv_image, boxes, class_ids, confidences
class YOLOV3_TINY: 
10class YOLOV3_TINY:
11    def __init__(self, class_names, model_path):
12        self.model = load_model(model_path)
13        self.class_names = class_names
14        self.timer = Timer()
15
16    def predict(self, image, top_k=-1, prob_threshold=None):
17        input_cv_image = cv2.resize(image, (416, 416))
18        input_cv_image = np.array(input_cv_image).astype(np.float32) / 255.0
19
20        # Yolo inference
21        self.timer.start()
22        outs = self.model.predict(np.array([input_cv_image]))
23        print("FPS: ", int(1.0 / self.timer.end()))
24
25        boxes, class_ids, confidences = yolo_process_output(outs, yolov3_tiny_anchors, len(self.class_names))
26
27        return input_cv_image, boxes, class_ids, confidences
YOLOV3_TINY(class_names, model_path) 
11    def __init__(self, class_names, model_path):
12        self.model = load_model(model_path)
13        self.class_names = class_names
14        self.timer = Timer()
def predict(self, image, top_k=-1, prob_threshold=None): 
16    def predict(self, image, top_k=-1, prob_threshold=None):
17        input_cv_image = cv2.resize(image, (416, 416))
18        input_cv_image = np.array(input_cv_image).astype(np.float32) / 255.0
19
20        # Yolo inference
21        self.timer.start()
22        outs = self.model.predict(np.array([input_cv_image]))
23        print("FPS: ", int(1.0 / self.timer.end()))
24
25        boxes, class_ids, confidences = yolo_process_output(outs, yolov3_tiny_anchors, len(self.class_names))
26
27        return input_cv_image, boxes, class_ids, confidences