import numpy as np import cv2 import pygame class main(): def __init__(self,traktor,field,ui,path): self.traktor = traktor self.field = field self.ui = ui self.path = path self.mode_value = 0 def get_output_layers(self,net): layer_names = net.getLayerNames() output_layers = [layer_names[i[0] - 1] for i in net.getUnconnectedOutLayers()] return output_layers def recognition(self,photo): image = photo Width = image.shape[1] Height = image.shape[0] scale = 0.00392 with open("si.names", 'r') as f: classes = [line.strip() for line in f.readlines()] COLORS = np.random.uniform(0, 255, size=(len(classes), 3)) net = cv2.dnn.readNet("si_20000.weights", "si.cfg") blob = cv2.dnn.blobFromImage(image, scale, (416, 416), (0, 0, 0), True, crop=False) net.setInput(blob) outs = net.forward(self.get_output_layers(net)) class_ids = [] confidences = [] boxes = [] conf_threshold = 0.5 nms_threshold = 0.4 for out in outs: for detection in out: scores = detection[5:] class_id = np.argmax(scores) confidence = scores[class_id] if confidence > 0.5: class_ids.append(class_id) return class_ids[0] def main(self): self.pole = self.ui.field_images[self.field.get_value(self.traktor.get_poz())] self.img = pygame.surfarray.array3d(self.pole) self.img = self.img.transpose([1,0,2]) self.img = cv2.cvtColor(self.img, cv2.COLOR_RGB2BGR) self.reco = self.mode(self.recognition(self.img)) if self.reco == 10: print("Nic nie trzeba robić") else: self.traktor.set_mode(self.reco) def mode(self,mode): self.mode_value = mode if self.mode_value in [0, 1, 2, 3]: return 0 elif self.mode_value in [1, 3, 5, 7]: return 1 elif self.mode_value in [0, 1, 4, 5]: return 2 elif self.mode_value in [8]: return 3 elif self.mode_value in [6]: return 10