def __init__(self, name_pattern, directory, frame_time, nimages=1, ntrigger=1, start_angle=0., angle_per_frame=0., image_nr_start=1, trigger_mode='exts', compression='bslz4'): self.detector = detector() self.beam_center = beam_center() self.name_pattern = name_pattern self.directory = directory self.frame_time = frame_time self.nimages = nimages self.ntrigger = ntrigger self.start_angle = start_angle self.angle_per_frame = angle_per_frame self.image_nr_start=1, self.trigger_mode = trigger_mode self.compression = compression
def __init__(self, name_pattern, directory, start, #dictionary of motor positions stop, #dictionary of motor positions vertical_range, #vertical single sweep range in milimiters frame_time=0.1, # in seconds number_of_points=None, # oscillation_start_angle=0, # angle in degrees total_oscillation_range=180, # angle in degrees degrees_per_frame=0.1, # angle per image in degrees degrees_of_overlap_between_neighboring_sweeps=1, # angle in degrees beam_horizontal_size=0.01): # beam width in mmm self.name_pattern = name_pattern self.directory = directory self.start = start self.stop = stop self.vertical_range = vertical_range self.beam_horizontal_size = beam_horizontal_size if number_of_points is None: start_vector = np.array([start[motor] for motor in self.motors]) stop_vector = np.array([stop[motor] for motor in self.motors]) scan_length = np.linalg.norm(stop_vector - start_vector) self.number_of_points = int(np.floor(scan_length/self.beam_horizontal_size)) print 'total length of principal helical line is %s' % scan_length print 'experiment will consist of %s vertical helical sweeps' % self.number_of_points self.oscillation_start_angle = oscillation_start_angle self.total_oscillation_range = total_oscillation_range self.degrees_per_frame = degrees_per_frame self.degrees_of_overlap_between_neighboring_sweeps = degrees_of_overlap_between_neighboring_sweeps self.frame_time = frame_time self.detector = detector() self.goniometer = goniometer()
while True: ret, frame = cap.read() if ret: start_time = time.time() frames = frame[:, :, ::-1] image = Image.fromarray(frames, 'RGB') width, high = image.size x_w = width / 416 y_h = high / 416 cropimg = image.resize((416, 416)) imgdata = transforms(cropimg) imgdata = torch.FloatTensor(imgdata).view(-1, 3, 416, 416).cuda() y = detector(imgdata, 0.5, cfg.ANCHORS_GROUP)[0] for i in y: x1 = int((i[0]) * x_w) y1 = int((i[1]) * y_h) x2 = int((i[2]) * x_w) y2 = int((i[3]) * y_h) cv2.rectangle(frame, (x1, y1), (x2, y2), (0, 0, 255)) end_time = time.time() print(end_time - start_time) cv2.imshow('a', frame) cv2.waitKey(0)
desp = get_feature(patch_gray, patch_grad, patch_dp, patchMask, ngray, ngrad, ndp, ps) keypoints.append([u, v]) features.append(desp) except: continue feats = scale(features, axis=0, with_mean=True, with_std=True) # normalize return keypoints, feats if __name__ == '__main__': import time from detector import * pathdir = './data/ManipulatorsDataset/' Kmat = np.loadtxt(pathdir + 'camera-intrinsics.txt') rgbfile = pathdir + 'mixture/illum/rgb_1.png' depthfile = pathdir + 'mixture/illum/depth_1.png' rgbImage = cv2.imread(rgbfile) grayImage = cv2.imread(rgbfile, 0) depthImage = cv2.imread(depthfile, cv2.IMREAD_UNCHANGED) kps = detector(grayImage, depthImage, Kmat) print(len(kps)) time1 = time.time() kps, desp = descriptor(grayImage, depthImage, kps, Kmat) print(time.time() - time1) print(len(kps))
def detector_runner(): global member_list, keep_running detective = detector(member_list) while (keep_running): time.sleep(2) detective.ping_neighbours()
import cfg if __name__ == '__main__': detector = Detector() data_for = transforms.Compose([transforms.ToTensor()]) cap = cv2.VideoCapture("D:/ai/ai/train2017/train2017/000000000394.jpg") while True: # ret 是否读取到图片 # 图片的数据 ret, frame = cap.read() if ret: # OPENCV读出来的数据是BGR格式,这里转换成RGB frames = frame[:, :, ::-1] image = Image.fromarray(frames, 'RGB') width, high = image.size x_w = width / 416 y_h = high / 416 cropimg = image.resize((416, 416)) imgdata = data_for(cropimg) imgdata = torch.FloatTensor(imgdata).view(-1, 3, 416, 416).cuda() y = detector(imgdata, 0.55, cfg.anchors_group)[0] print(y) for i in y: x1 = int((i[0]) * x_w) y1 = int((i[1]) * y_h) x2 = int((i[2]) * x_w) y2 = int((i[3]) * y_h) cv2.rectangle(frame, (x1, y1), (x2, y2), (0, 0, 255)) cv2.imshow('a', frame) cv2.waitKey(0)
if __name__ == '__main__': camera = cv2.VideoCapture(video_filename) # Creates graph from saved GraphDef. create_graph() while (1): start = time.time() # want to time each cycle. starting stopwatch. num_throw_away_frames = 30 # simulate speed of classifier in real-time situation for i in xrange(num_throw_away_frames): retval, temp = camera.read() retval, camera_capture = camera.read() file = "test_image.png" cv2.imwrite(file, camera_capture) vote, target_label = detector('test_image.png') print(target_label + ' is present.') end = time.time() print('Elapsed Time: ' + str(end - start)) # print out how long this detection cycle took print('********') ### uncomment this section if you want to see the images that ### detector classified. if you do invoke this section, loop will ### pause on each image till keystroke, ESC will quit the program #if(vote): # cv2.circle(camera_capture,(320, 240), 50, (0,255,0), 5) #cv2.imshow('image', camera_capture) #k = cv2.waitKey(0)