Я прибег к использованию процесса обучения в облаке. Учитывая продукт, который я получил, я бы ожидал, что перейду непосредственно к имеющемуся у меня коду, который работает с другими моделями tflite, но модель, созданная в облаке, не работает. Я получаю «индекс вне диапазона» при запросе interpreter.get_tensor
параметров.
Вот мой код, в основном модифицированный пример, где я могу принять видео и создать видео с результатами.
import argparse
import cv2
import numpy as np
import sys
import importlib.util
# Define and parse input arguments
parser = argparse.ArgumentParser()
parser.add_argument('--modeldir', help='Folder the .tflite file is located in',
required=True)
parser.add_argument('--graph', help='Name of the .tflite file, if different than detect.tflite',
default='model.tflite')
# default='/tmp/detect.tflite')
parser.add_argument('--labels', help='Name of the labelmap file, if different than labelmap.txt',
default='dict.txt')
# default='/tmp/coco_labels.txt')
parser.add_argument('--threshold', help='Minimum confidence threshold for displaying detected objects',
default=0.5)
parser.add_argument('--video', help='Name of the video file',
default='test.mp4')
parser.add_argument('--edgetpu', help='Use Coral Edge TPU Accelerator to speed up detection',
action='store_true')
args = parser.parse_args()
MODEL_NAME = args.modeldir
GRAPH_NAME = args.graph
LABELMAP_NAME = args.labels
VIDEO_NAME = args.video
min_conf_threshold = float(args.threshold)
use_TPU = args.edgetpu
# Import TensorFlow libraries
# If tensorflow is not installed, import interpreter from tflite_runtime, else import from regular tensorflow
# If using Coral Edge TPU, import the load_delegate library
pkg = importlib.util.find_spec('tensorflow')
pkg = True
if pkg is None:
from tflite_runtime.interpreter import Interpreter
if use_TPU:
from tflite_runtime.interpreter import load_delegate
else:
from tensorflow.lite.python.interpreter import Interpreter
if use_TPU:
from tensorflow.lite.python.interpreter import load_delegate
# If using Edge TPU, assign filename for Edge TPU model
if use_TPU:
# If user has specified the name of the .tflite file, use that name, otherwise use default 'edgetpu.tflite'
if (GRAPH_NAME == 'detect.tflite'):
GRAPH_NAME = 'edgetpu.tflite'
# Get path to current working directory
CWD_PATH = os.getcwd()
# Path to video file
VIDEO_PATH = os.path.join(CWD_PATH,VIDEO_NAME)
# Path to .tflite file, which contains the model that is used for object detection
PATH_TO_CKPT = os.path.join(CWD_PATH,MODEL_NAME,GRAPH_NAME)
# Path to label map file
PATH_TO_LABELS = os.path.join(CWD_PATH,MODEL_NAME,LABELMAP_NAME)
# Load the label map
with open(PATH_TO_LABELS, 'r') as f:
labels = [line.strip() for line in f.readlines()]
# Have to do a weird fix for label map if using the COCO "starter model" from
# https://www.tensorflow.org/lite/models/object_detection/overview
# First label is '???', which has to be removed.
if labels[0] == '???':
del(labels[0])
# Load the Tensorflow Lite model.
# If using Edge TPU, use special load_delegate argument
if use_TPU:
interpreter = Interpreter(model_path=PATH_TO_CKPT,
experimental_delegates=[load_delegate('libedgetpu.so.1.0')])
print(PATH_TO_CKPT)
else:
interpreter = Interpreter(model_path=PATH_TO_CKPT)
interpreter.allocate_tensors()
# Get model details
input_details = interpreter.get_input_details()
output_details = interpreter.get_output_details()
height = input_details[0]['shape'][1]
width = input_details[0]['shape'][2]
floating_model = (input_details[0]['dtype'] == np.float32)
input_mean = 127.5
input_std = 127.5
# Open video file
video = cv2.VideoCapture(VIDEO_PATH)
imW = video.get(cv2.CAP_PROP_FRAME_WIDTH)
imH = video.get(cv2.CAP_PROP_FRAME_HEIGHT)
out = cv2.VideoWriter('output.avi', cv2.VideoWriter_fourcc(
'M', 'J', 'P', 'G'), 10, (1920, 1080))
while(video.isOpened()):
# Acquire frame and resize to expected shape [1xHxWx3]
ret, frame = video.read()
frame_rgb = cv2.cvtColor(frame, cv2.COLOR_BGR2RGB)
frame_resized = cv2.resize(frame_rgb, (width, height))
input_data = np.expand_dims(frame_resized, axis=0)
# Normalize pixel values if using a floating model (i.e. if model is non-quantized)
if floating_model:
input_data = (np.float32(input_data) - input_mean) / input_std
# Perform the actual detection by running the model with the image as input
interpreter.set_tensor(input_details[0]['index'],input_data)
interpreter.invoke()
# Retrieve detection results
boxes = interpreter.get_tensor(output_details[0]['index'])[0] # Bounding box coordinates of detected objects
classes = interpreter.get_tensor(output_details[1]['index'])[0] # Class index of detected objects
scores = interpreter.get_tensor(output_details[2]['index'])[0] # Confidence of detected objects
print (boxes)
print (classes)
print (scores)
#num = interpreter.get_tensor(output_details[3]['index'])[0] # Total number of detected objects (inaccurate and not needed)
# Loop over all detections and draw detection box if confidence is above minimum threshold
for i in range(len(scores)):
if ((scores[i] > min_conf_threshold) and (scores[i] <= 1.0)):
# Get bounding box coordinates and draw box
# Interpreter can return coordinates that are outside of image dimensions, need to force them to be within image using max() and min()
ymin = int(max(1,(boxes[i][0] * imH)))
xmin = int(max(1,(boxes[i][1] * imW)))
ymax = int(min(imH,(boxes[i][2] * imH)))
xmax = int(min(imW,(boxes[i][3] * imW)))
cv2.rectangle(frame, (xmin,ymin), (xmax,ymax), (10, 255, 0), 4)
# Draw label
object_name = labels[int(classes[i])] # Look up object name from "labels" array using class index
label = '%s: %d%%' % (object_name, int(scores[i]*100)) # Example: 'person: 72%'
labelSize, baseLine = cv2.getTextSize(label, cv2.FONT_HERSHEY_SIMPLEX, 0.7, 2) # Get font size
label_ymin = max(ymin, labelSize[1] + 10) # Make sure not to draw label too close to top of window
cv2.rectangle(frame, (xmin, label_ymin-labelSize[1]-10), (xmin+labelSize[0],
label_ymin+baseLine-10), (255, 255, 255), cv2.FILLED) # Draw white box to put label text in
cv2.putText(frame, label, (xmin, label_ymin-7), cv2.FONT_HERSHEY_SIMPLEX,
0.7, (0, 0, 0), 2) # Draw label text
# All the results have been drawn on the frame, so it's time to display it.
cv2.imshow('Object detector', frame)
#output_rgb = cv2.cvtColor(frame, cv2.COLOR_RGB2BGR)
out.write(frame)
# Press 'q' to quit
if cv2.waitKey(1) == ord('q'):
break
# Clean up
video.release()
out.release()
cv2.destroyAllWindows()
Вот как должны выглядеть операторы печати при использовании стандартной модели tflite:
[32. 76. 56. 76. 0. 61. 74. 0. 0. 0.]
[0.609375 0.48828125 0.44921875 0.44921875 0.4140625 0.40234375
0.37890625 0.3125 0.3125 0.3125 ]
[[-0.01923192 0.17330796 0.747546 0.8384144 ]
[ 0.01866053 0.5023282 0.39603746 0.6143299 ]
[ 0.01673795 0.47382414 0.34407628 0.5580931 ]
[ 0.11588445 0.78543806 0.8778869 1.0039229 ]
[ 0.8106107 0.70675755 1.0080075 0.89248717]
[ 0.84941524 0.06391776 1.0006479 0.28792098]
[ 0.05543692 0.53557926 0.40413857 0.62823087]
[ 0.07051808 -0.00938512 0.8822515 0.28100258]
[ 0.68205094 0.33990026 0.9940187 0.6020821 ]
[ 0.08010477 0.01998334 0.6011186 0.26135433]]
Вот ошибка при представлении модели, созданной в облаке:
File "tflite_vid.py", line 124, in <module>
classes = interpreter.get_tensor(output_details[1]['index'])[0] # Class index of detected objects
IndexError: list index out of range
Поэтому я хотел бы попросить кого-нибудь объяснить, как разработать модель TFlite с TF2 с Python или как получить облако для создания пригодной модели TFlite. Пожалуйста, о, пожалуйста, не указывайте мне направление, которое влечет за собой удивление на примерах Inte rnet, если только они не являются настоящим Евангелием о том, как это сделать.,