Shuyao "Tim" Xu
Introduce Preprocessing for Optimized Quantization in `quantize-ort.py` (#238)
2129055
# This file is part of OpenCV Zoo project. | |
# It is subject to the license terms in the LICENSE file found in the same directory. | |
# | |
# Copyright (C) 2021, Shenzhen Institute of Artificial Intelligence and Robotics for Society, all rights reserved. | |
# Third party copyrights are property of their respective owners. | |
import os | |
import sys | |
import numpy as np | |
import cv2 as cv | |
import onnx | |
from onnx import version_converter | |
import onnxruntime | |
from onnxruntime.quantization import quantize_static, CalibrationDataReader, QuantType, QuantFormat, quant_pre_process | |
from transform import Compose, Resize, CenterCrop, Normalize, ColorConvert, HandAlign | |
class DataReader(CalibrationDataReader): | |
def __init__(self, model_path, image_dir, transforms, data_dim): | |
model = onnx.load(model_path) | |
self.input_name = model.graph.input[0].name | |
self.transforms = transforms | |
self.data_dim = data_dim | |
self.data = self.get_calibration_data(image_dir) | |
self.enum_data_dicts = iter([{self.input_name: x} for x in self.data]) | |
def get_next(self): | |
return next(self.enum_data_dicts, None) | |
def get_calibration_data(self, image_dir): | |
blobs = [] | |
supported = ["jpg", "png"] # supported file suffix | |
for image_name in os.listdir(image_dir): | |
image_name_suffix = image_name.split('.')[-1].lower() | |
if image_name_suffix not in supported: | |
continue | |
img = cv.imread(os.path.join(image_dir, image_name)) | |
img = self.transforms(img) | |
if img is None: | |
continue | |
blob = cv.dnn.blobFromImage(img) | |
if self.data_dim == 'hwc': | |
blob = cv.transposeND(blob, [0, 2, 3, 1]) | |
blobs.append(blob) | |
return blobs | |
class Quantize: | |
def __init__(self, model_path, calibration_image_dir, transforms=Compose(), per_channel=False, act_type='int8', wt_type='int8', data_dim='chw', nodes_to_exclude=[]): | |
self.type_dict = {"uint8" : QuantType.QUInt8, "int8" : QuantType.QInt8} | |
self.model_path = model_path | |
self.calibration_image_dir = calibration_image_dir | |
self.transforms = transforms | |
self.per_channel = per_channel | |
self.act_type = act_type | |
self.wt_type = wt_type | |
self.nodes_to_exclude = nodes_to_exclude | |
# data reader | |
self.dr = DataReader(self.model_path, self.calibration_image_dir, self.transforms, data_dim) | |
def check_opset(self): | |
model = onnx.load(self.model_path) | |
if model.opset_import[0].version != 13: | |
print('\tmodel opset version: {}. Converting to opset 13'.format(model.opset_import[0].version)) | |
# convert opset version to 13 | |
model_opset13 = version_converter.convert_version(model, 13) | |
# save converted model | |
output_name = '{}-opset13.onnx'.format(self.model_path[:-5]) | |
onnx.save_model(model_opset13, output_name) | |
# update model_path for quantization | |
return output_name | |
return self.model_path | |
def run(self): | |
print('Quantizing {}: act_type {}, wt_type {}'.format(self.model_path, self.act_type, self.wt_type)) | |
new_model_path = self.check_opset() | |
quant_pre_process(new_model_path, new_model_path) | |
output_name = '{}_{}.onnx'.format(self.model_path[:-5], self.wt_type) | |
quantize_static(new_model_path, output_name, self.dr, | |
quant_format=QuantFormat.QOperator, # start from onnxruntime==1.11.0, quant_format is set to QuantFormat.QDQ by default, which performs fake quantization | |
per_channel=self.per_channel, | |
weight_type=self.type_dict[self.wt_type], | |
activation_type=self.type_dict[self.act_type], | |
nodes_to_exclude=self.nodes_to_exclude) | |
if new_model_path != self.model_path: | |
os.remove(new_model_path) | |
print('\tQuantized model saved to {}'.format(output_name)) | |
models=dict( | |
yunet=Quantize(model_path='../../models/face_detection_yunet/face_detection_yunet_2023mar.onnx', | |
calibration_image_dir='../../benchmark/data/face_detection', | |
transforms=Compose([Resize(size=(160, 120))]), | |
nodes_to_exclude=['MaxPool_5', 'MaxPool_18', 'MaxPool_25', 'MaxPool_32'], | |
), | |
sface=Quantize(model_path='../../models/face_recognition_sface/face_recognition_sface_2021dec.onnx', | |
calibration_image_dir='../../benchmark/data/face_recognition', | |
transforms=Compose([Resize(size=(112, 112))])), | |
pphumanseg=Quantize(model_path='../../models/human_segmentation_pphumanseg/human_segmentation_pphumanseg_2023mar.onnx', | |
calibration_image_dir='../../benchmark/data/human_segmentation', | |
transforms=Compose([Resize(size=(192, 192))])), | |
ppresnet50=Quantize(model_path='../../models/image_classification_ppresnet/image_classification_ppresnet50_2022jan.onnx', | |
calibration_image_dir='../../benchmark/data/image_classification', | |
transforms=Compose([Resize(size=(224, 224))])), | |
# TBD: VitTrack | |
youtureid=Quantize(model_path='../../models/person_reid_youtureid/person_reid_youtu_2021nov.onnx', | |
calibration_image_dir='../../benchmark/data/person_reid', | |
transforms=Compose([Resize(size=(128, 256))])), | |
ppocrv3det_en=Quantize(model_path='../../models/text_detection_ppocr/text_detection_en_ppocrv3_2023may.onnx', | |
calibration_image_dir='../../benchmark/data/text', | |
transforms=Compose([Resize(size=(736, 736)), | |
Normalize(mean=[123.675, 116.28, 103.53], std=[58.395, 57.12, 57.375])])), | |
ppocrv3det_cn=Quantize(model_path='../../models/text_detection_ppocr/text_detection_cn_ppocrv3_2023may.onnx', | |
calibration_image_dir='../../benchmark/data/text', | |
transforms=Compose([Resize(size=(736, 736)), | |
Normalize(mean=[123.675, 116.28, 103.53], std=[58.395, 57.12, 57.375])])), | |
crnn_en=Quantize(model_path='../../models/text_recognition_crnn/text_recognition_CRNN_EN_2021sep.onnx', | |
calibration_image_dir='../../benchmark/data/text', | |
transforms=Compose([Resize(size=(100, 32)), Normalize(mean=[127.5, 127.5, 127.5], std=[127.5, 127.5, 127.5]), ColorConvert(ctype=cv.COLOR_BGR2GRAY)])), | |
crnn_cn=Quantize(model_path='../../models/text_recognition_crnn/text_recognition_CRNN_CN_2021nov.onnx', | |
calibration_image_dir='../../benchmark/data/text', | |
transforms=Compose([Resize(size=(100, 32))])), | |
mp_palmdet=Quantize(model_path='../../models/palm_detection_mediapipe/palm_detection_mediapipe_2023feb.onnx', | |
calibration_image_dir='path/to/dataset', | |
transforms=Compose([Resize(size=(192, 192)), Normalize(std=[255, 255, 255]), | |
ColorConvert(ctype=cv.COLOR_BGR2RGB)]), data_dim='hwc'), | |
mp_handpose=Quantize(model_path='../../models/handpose_estimation_mediapipe/handpose_estimation_mediapipe_2023feb.onnx', | |
calibration_image_dir='path/to/dataset', | |
transforms=Compose([HandAlign("mp_handpose"), Resize(size=(224, 224)), Normalize(std=[255, 255, 255]), | |
ColorConvert(ctype=cv.COLOR_BGR2RGB)]), data_dim='hwc'), | |
lpd_yunet=Quantize(model_path='../../models/license_plate_detection_yunet/license_plate_detection_lpd_yunet_2023mar.onnx', | |
calibration_image_dir='../../benchmark/data/license_plate_detection', | |
transforms=Compose([Resize(size=(320, 240))]), | |
nodes_to_exclude=['MaxPool_5', 'MaxPool_18', 'MaxPool_25', 'MaxPool_32', 'MaxPool_39'], | |
), | |
) | |
if __name__ == '__main__': | |
selected_models = [] | |
for i in range(1, len(sys.argv)): | |
selected_models.append(sys.argv[i]) | |
if not selected_models: | |
selected_models = list(models.keys()) | |
print('Models to be quantized: {}'.format(str(selected_models))) | |
for selected_model_name in selected_models: | |
q = models[selected_model_name] | |
q.run() | |