Deploy yolov5 with OpenVINO

deploy yolov5 with OpenVINO, mainly in python, while C++ to be solved

Environment

• Windows 10
• CPU Intel i9-10900k
• OpenVINO 2021.4
• Python 3.6.9
• ONNX 1.8.1
• Pytorch 1.10.0
• Netron 5.3.1
• Yolov5 5.0

 

Install OpenVINO

Download OpenVINO

  What is OpenVINO
  Download OpenVINO 2021.4

 

Setting environment variables

  Before OpenVINO can be compiled and run, several environment variables must be updated. Open the command prompt and run the following batch file to temporarily set the environment variables:

C:\Program Files(x86)\IntelSWTools\openvino\bin\setupvars.bat

  Add system environment variables as well

F:\openvino\openvino_2021.4.689\deployment_tools\ngraph\lib
F:\openvino\openvino_2021.4.689\deployment_tools\inference_engine\bin\intel64\Release
F:\openvino\openvino_2021.4.689\deployment_tools\inference_engine\external\tbb\bin
F:\openvino\openvino_2021.4.689\opencv\bin

  When somrthing wrong with the installation of openvino

 

Convert Pytorch Weights to ONNX Weights

Modify models/common.py

  Find the 39th line in models/common.py shown as follows

self.act = nn.SiLU() if act is True else (act if isinstance(act, nn.Module) else nn.Identity())

  Change it to the following

self.act = nn.LeakyReLU(0.1, inplace=True)

 

Modify models/export.py

  Modify the parameters from the 23th line to 29th line in models/export.py, especially the weight you want to convert

  Revise the script models/export.py in line 77 to opset version 10

 

Export onnx weight

  Then run export.py or run following script, and we get weight.onnx

python3 models/export.py  --weights weight.pt  --img-size 640

 

Solutions with export errors

• When there is some error with coremltools, ensure your python's version is earlier than 3.7, and you'd better start a new virtual environment
• When it shows "ModuleNotFoundError: No module named 'utils'" , you should execute script "pip install utils"
• When it shows "ModuleNotFoundError: No module named 'utils.google_utils'", then see
• When it shows "pytorch Key already registered with the same priority: GroupSpatialSoftmax", then reinstall pytorch

 

Convert ONNX weight to IR File

  Before you convert the weight, you need to confirm that you have set OpenVINO environment and variables

Find the output nodes

  We need to specify the output node of the IR when we use model optimizer to convert the ONNX weights file. For example, there are 3 output nodes in yolov5s.onnx. We can use Netron to visualize yolov5s.onnx. Then we find the output nodes by searching the keyword “Transpose” in Netron.

  After that, we can find the convolution node marked as oval shown in following Figure. After double clicking the convolution node, we can read its name “Conv_249”. Repeat the above steps to discover the other two nodes named “Conv_243” and "Conv_246".

Transfer ONNX weight to IR

  Use following script to get IR file, then we will get IR files(weight.xml and weight.bin) of weights

python F:\\openvino\\openvino_2021.4.689\\deployment_tools\\model_o
ptimizer\\mo.py  --input_model weights/yolov5s.onnx -s 255 --reverse_input_channels --output Conv_243,Conv_246,Conv_249

 

Inference with OpenVINO

Establish python file

  Establish yolov5_demo_OV2021.3.py with following content

#!/usr/bin/env python
"""
 Copyright (C) 2018-2019 Intel Corporation

 Licensed under the Apache License, Version 2.0 (the "License");
 you may not use this file except in compliance with the License.
 You may obtain a copy of the License at

      http://www.apache.org/licenses/LICENSE-2.0

 Unless required by applicable law or agreed to in writing, software
 distributed under the License is distributed on an "AS IS" BASIS,
 WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
 See the License for the specific language governing permissions and
 limitations under the License.
"""
from __future__ import print_function, division

import logging
import os
import sys
from argparse import ArgumentParser, SUPPRESS
from math import exp as exp
from time import time
import numpy as np

import cv2
from openvino.inference_engine import IENetwork, IECore

logging.basicConfig(format="[ %(levelname)s ] %(message)s", level=logging.INFO, stream=sys.stdout)
log = logging.getLogger()


def build_argparser():
    parser = ArgumentParser(add_help=False)
    args = parser.add_argument_group('Options')
    args.add_argument('-h', '--help', action='help', default=SUPPRESS, help='Show this help message and exit.')
    args.add_argument("-m", "--model", help="Required. Path to an .xml file with a trained model.",
                      required=True, type=str)
    args.add_argument("-i", "--input", help="Required. Path to an image/video file. (Specify 'cam' to work with "
                                            "camera)", required=True, type=str)
    args.add_argument("-l", "--cpu_extension",
                      help="Optional. Required for CPU custom layers. Absolute path to a shared library with "
                           "the kernels implementations.", type=str, default=None)
    args.add_argument("-d", "--device",
                      help="Optional. Specify the target device to infer on; CPU, GPU, FPGA, HDDL or MYRIAD is"
                           " acceptable. The sample will look for a suitable plugin for device specified. "
                           "Default value is CPU", default="CPU", type=str)
    args.add_argument("--labels", help="Optional. Labels mapping file", default=None, type=str)
    args.add_argument("-t", "--prob_threshold", help="Optional. Probability threshold for detections filtering",
                      default=0.5, type=float)
    args.add_argument("-iout", "--iou_threshold", help="Optional. Intersection over union threshold for overlapping "
                                                       "detections filtering", default=0.4, type=float)
    args.add_argument("-ni", "--number_iter", help="Optional. Number of inference iterations", default=1, type=int)
    args.add_argument("-pc", "--perf_counts", help="Optional. Report performance counters", default=False,
                      action="store_true")
    args.add_argument("-r", "--raw_output_message", help="Optional. Output inference results raw values showing",
                      default=False, action="store_true")
    args.add_argument("--no_show", help="Optional. Don't show output", action='store_true')
    return parser


class YoloParams:
    # ------------------------------------------- Extracting layer parameters ------------------------------------------
    # Magic numbers are copied from yolo samples
    def __init__(self,  side):
        self.num = 3 #if 'num' not in param else int(param['num'])
        self.coords = 4 #if 'coords' not in param else int(param['coords'])
        self.classes = 80 #if 'classes' not in param else int(param['classes'])
        self.side = side
        self.anchors = [10.0, 13.0, 16.0, 30.0, 33.0, 23.0, 30.0, 61.0, 62.0, 45.0, 59.0, 119.0, 116.0, 90.0, 156.0,
                        198.0,
                        373.0, 326.0] #if 'anchors' not in param else [float(a) for a in param['anchors'].split(',')]

        #self.isYoloV3 = False

        #if param.get('mask'):
        #    mask = [int(idx) for idx in param['mask'].split(',')]
        #    self.num = len(mask)

        #    maskedAnchors = []
        #    for idx in mask:
        #        maskedAnchors += [self.anchors[idx * 2], self.anchors[idx * 2 + 1]]
        #    self.anchors = maskedAnchors

        #    self.isYoloV3 = True # Weak way to determine but the only one.

    def log_params(self):
        params_to_print = {'classes': self.classes, 'num': self.num, 'coords': self.coords, 'anchors': self.anchors}
        [log.info("         {:8}: {}".format(param_name, param)) for param_name, param in params_to_print.items()]


def letterbox(img, size=(640, 640), color=(114, 114, 114), auto=True, scaleFill=False, scaleup=True):
    # Resize image to a 32-pixel-multiple rectangle https://github.com/ultralytics/yolov3/issues/232
    shape = img.shape[:2]  # current shape [height, width]
    w, h = size

    # Scale ratio (new / old)
    r = min(h / shape[0], w / shape[1])
    if not scaleup:  # only scale down, do not scale up (for better test mAP)
        r = min(r, 1.0)

    # Compute padding
    ratio = r, r  # width, height ratios
    new_unpad = int(round(shape[1] * r)), int(round(shape[0] * r))
    dw, dh = w - new_unpad[0], h - new_unpad[1]  # wh padding
    if auto:  # minimum rectangle
        dw, dh = np.mod(dw, 64), np.mod(dh, 64)  # wh padding
    elif scaleFill:  # stretch
        dw, dh = 0.0, 0.0
        new_unpad = (w, h)
        ratio = w / shape[1], h / shape[0]  # width, height ratios

    dw /= 2  # divide padding into 2 sides
    dh /= 2

    if shape[::-1] != new_unpad:  # resize
        img = cv2.resize(img, new_unpad, interpolation=cv2.INTER_LINEAR)
    top, bottom = int(round(dh - 0.1)), int(round(dh + 0.1))
    left, right = int(round(dw - 0.1)), int(round(dw + 0.1))
    img = cv2.copyMakeBorder(img, top, bottom, left, right, cv2.BORDER_CONSTANT, value=color)  # add border

    top2, bottom2, left2, right2 = 0, 0, 0, 0
    if img.shape[0] != h:
        top2 = (h - img.shape[0])//2
        bottom2 = top2
        img = cv2.copyMakeBorder(img, top2, bottom2, left2, right2, cv2.BORDER_CONSTANT, value=color)  # add border
    elif img.shape[1] != w:
        left2 = (w - img.shape[1])//2
        right2 = left2
        img = cv2.copyMakeBorder(img, top2, bottom2, left2, right2, cv2.BORDER_CONSTANT, value=color)  # add border
    return img


def scale_bbox(x, y, height, width, class_id, confidence, im_h, im_w, resized_im_h=640, resized_im_w=640):
    gain = min(resized_im_w / im_w, resized_im_h / im_h)  # gain  = old / new
    pad = (resized_im_w - im_w * gain) / 2, (resized_im_h - im_h * gain) / 2  # wh padding
    x = int((x - pad[0])/gain)
    y = int((y - pad[1])/gain)

    w = int(width/gain)
    h = int(height/gain)
 
    xmin = max(0, int(x - w / 2))
    ymin = max(0, int(y - h / 2))
    xmax = min(im_w, int(xmin + w))
    ymax = min(im_h, int(ymin + h))
    # Method item() used here to convert NumPy types to native types for compatibility with functions, which don't
    # support Numpy types (e.g., cv2.rectangle doesn't support int64 in color parameter)
    return dict(xmin=xmin, xmax=xmax, ymin=ymin, ymax=ymax, class_id=class_id.item(), confidence=confidence.item())


def entry_index(side, coord, classes, location, entry):
    side_power_2 = side ** 2
    n = location // side_power_2
    loc = location % side_power_2
    return int(side_power_2 * (n * (coord + classes + 1) + entry) + loc)


def parse_yolo_region(blob, resized_image_shape, original_im_shape, params, threshold):
    # ------------------------------------------ Validating output parameters ------------------------------------------    
    out_blob_n, out_blob_c, out_blob_h, out_blob_w = blob.shape
    predictions = 1.0/(1.0+np.exp(-blob)) 
                   
    assert out_blob_w == out_blob_h, "Invalid size of output blob. It sould be in NCHW layout and height should " \
                                     "be equal to width. Current height = {}, current width = {}" \
                                     "".format(out_blob_h, out_blob_w)

    # ------------------------------------------ Extracting layer parameters -------------------------------------------
    orig_im_h, orig_im_w = original_im_shape
    resized_image_h, resized_image_w = resized_image_shape
    objects = list()
 
    side_square = params.side * params.side

    # ------------------------------------------- Parsing YOLO Region output -------------------------------------------
    bbox_size = int(out_blob_c/params.num) #4+1+num_classes

    for row, col, n in np.ndindex(params.side, params.side, params.num):
        bbox = predictions[0, n*bbox_size:(n+1)*bbox_size, row, col]
        
        x, y, width, height, object_probability = bbox[:5]
        class_probabilities = bbox[5:]
        if object_probability < threshold:
            continue
        x = (2*x - 0.5 + col)*(resized_image_w/out_blob_w)
        y = (2*y - 0.5 + row)*(resized_image_h/out_blob_h)
        if int(resized_image_w/out_blob_w) == 8 & int(resized_image_h/out_blob_h) == 8: #80x80, 
            idx = 0
        elif int(resized_image_w/out_blob_w) == 16 & int(resized_image_h/out_blob_h) == 16: #40x40
            idx = 1
        elif int(resized_image_w/out_blob_w) == 32 & int(resized_image_h/out_blob_h) == 32: # 20x20
            idx = 2

        width = (2*width)**2* params.anchors[idx * 6 + 2 * n]
        height = (2*height)**2 * params.anchors[idx * 6 + 2 * n + 1]
        class_id = np.argmax(class_probabilities)
        confidence = object_probability
        objects.append(scale_bbox(x=x, y=y, height=height, width=width, class_id=class_id, confidence=confidence,
                                  im_h=orig_im_h, im_w=orig_im_w, resized_im_h=resized_image_h, resized_im_w=resized_image_w))
    return objects


def intersection_over_union(box_1, box_2):
    width_of_overlap_area = min(box_1['xmax'], box_2['xmax']) - max(box_1['xmin'], box_2['xmin'])
    height_of_overlap_area = min(box_1['ymax'], box_2['ymax']) - max(box_1['ymin'], box_2['ymin'])
    if width_of_overlap_area < 0 or height_of_overlap_area < 0:
        area_of_overlap = 0
    else:
        area_of_overlap = width_of_overlap_area * height_of_overlap_area
    box_1_area = (box_1['ymax'] - box_1['ymin']) * (box_1['xmax'] - box_1['xmin'])
    box_2_area = (box_2['ymax'] - box_2['ymin']) * (box_2['xmax'] - box_2['xmin'])
    area_of_union = box_1_area + box_2_area - area_of_overlap
    if area_of_union == 0:
        return 0
    return area_of_overlap / area_of_union


def main():
    args = build_argparser().parse_args()


    # ------------- 1. Plugin initialization for specified device and load extensions library if specified -------------
    log.info("Creating Inference Engine...")
    ie = IECore()
    if args.cpu_extension and 'CPU' in args.device:
        ie.add_extension(args.cpu_extension, "CPU")

    # -------------------- 2. Reading the IR generated by the Model Optimizer (.xml and .bin files) --------------------
    model = args.model
    log.info(f"Loading network:\n\t{model}")
    net = ie.read_network(model=model)

    # ---------------------------------- 3. Load CPU extension for support specific layer ------------------------------
#    if "CPU" in args.device:
#        supported_layers = ie.query_network(net, "CPU")
#        not_supported_layers = [l for l in net.layers.keys() if l not in supported_layers]
#        if len(not_supported_layers) != 0:
#            log.error("Following layers are not supported by the plugin for specified device {}:\n {}".
#                     format(args.device, ', '.join(not_supported_layers)))
#            log.error("Please try to specify cpu extensions library path in sample's command line parameters using -l "
#                      "or --cpu_extension command line argument")
#            sys.exit(1)

    assert len(net.input_info.keys()) == 1, "Sample supports only YOLO V3 based single input topologies"

    # ---------------------------------------------- 4. Preparing inputs -----------------------------------------------
    log.info("Preparing inputs")
    input_blob = next(iter(net.input_info))

    #  Defaulf batch_size is 1
    net.batch_size = 1

    # Read and pre-process input images
    n, c, h, w = net.input_info[input_blob].input_data.shape

    if args.labels:
        with open(args.labels, 'r') as f:
            labels_map = [x.strip() for x in f]
    else:
        labels_map = None

    input_stream = 0 if args.input == "cam" else args.input

    is_async_mode = True
    cap = cv2.VideoCapture(input_stream)
    number_input_frames = int(cap.get(cv2.CAP_PROP_FRAME_COUNT))
    number_input_frames = 1 if number_input_frames != -1 and number_input_frames < 0 else number_input_frames

    wait_key_code = 1

    # Number of frames in picture is 1 and this will be read in cycle. Sync mode is default value for this case
    if number_input_frames != 1:
        ret, frame = cap.read()
    else:
        is_async_mode = False
        wait_key_code = 0

    # ----------------------------------------- 5. Loading model to the plugin -----------------------------------------
    log.info("Loading model to the plugin")
    exec_net = ie.load_network(network=net, num_requests=2, device_name=args.device)

    cur_request_id = 0
    next_request_id = 1
    render_time = 0
    parsing_time = 0

    # ----------------------------------------------- 6. Doing inference -----------------------------------------------
    log.info("Starting inference...")
    print("To close the application, press 'CTRL+C' here or switch to the output window and press ESC key")
    print("To switch between sync/async modes, press TAB key in the output window")
    while cap.isOpened():
        # Here is the first asynchronous point: in the Async mode, we capture frame to populate the NEXT infer request
        # in the regular mode, we capture frame to the CURRENT infer request
        if is_async_mode:
            ret, next_frame = cap.read()
        else:
            ret, frame = cap.read()

        if not ret:
            break

        if is_async_mode:
            request_id = next_request_id
            in_frame = letterbox(frame, (w, h))
        else:
            request_id = cur_request_id
            in_frame = letterbox(frame, (w, h))

        in_frame0 = in_frame
        # resize input_frame to network size
        in_frame = in_frame.transpose((2, 0, 1))  # Change data layout from HWC to CHW
        in_frame = in_frame.reshape((n, c, h, w))

        # Start inference
        start_time = time()
        exec_net.start_async(request_id=request_id, inputs={input_blob: in_frame})
        det_time = time() - start_time

        # Collecting object detection results
        objects = list()
        if exec_net.requests[cur_request_id].wait(-1) == 0:
            output = exec_net.requests[cur_request_id].output_blobs
            start_time = time()
            for layer_name, out_blob in output.items():
                layer_params = YoloParams(side=out_blob.buffer.shape[2])
                log.info("Layer {} parameters: ".format(layer_name))
                layer_params.log_params()
                objects += parse_yolo_region(out_blob.buffer, in_frame.shape[2:],
                                             #in_frame.shape[2:], layer_params,
                                             frame.shape[:-1], layer_params,
                                             args.prob_threshold)
            parsing_time = time() - start_time

        # Filtering overlapping boxes with respect to the --iou_threshold CLI parameter
        objects = sorted(objects, key=lambda obj : obj['confidence'], reverse=True)
        for i in range(len(objects)):
            if objects[i]['confidence'] == 0:
                continue
            for j in range(i + 1, len(objects)):
                if intersection_over_union(objects[i], objects[j]) > args.iou_threshold:
                    objects[j]['confidence'] = 0

        # Drawing objects with respect to the --prob_threshold CLI parameter
        objects = [obj for obj in objects if obj['confidence'] >= args.prob_threshold]

        if len(objects) and args.raw_output_message:
            log.info("\nDetected boxes for batch {}:".format(1))
            log.info(" Class ID | Confidence | XMIN | YMIN | XMAX | YMAX | COLOR ")

        origin_im_size = frame.shape[:-1]
        print(origin_im_size)
        for obj in objects:
            # Validation bbox of detected object
            if obj['xmax'] > origin_im_size[1] or obj['ymax'] > origin_im_size[0] or obj['xmin'] < 0 or obj['ymin'] < 0:
                continue
            color = (int(min(obj['class_id'] * 12.5, 255)),
                     min(obj['class_id'] * 7, 255), min(obj['class_id'] * 5, 255))
            det_label = labels_map[obj['class_id']] if labels_map and len(labels_map) >= obj['class_id'] else \
                str(obj['class_id'])

            if args.raw_output_message:
                log.info(
                    "{:^9} | {:10f} | {:4} | {:4} | {:4} | {:4} | {} ".format(det_label, obj['confidence'], obj['xmin'],
                                                                              obj['ymin'], obj['xmax'], obj['ymax'],
                                                                              color))

            cv2.rectangle(frame, (obj['xmin'], obj['ymin']), (obj['xmax'], obj['ymax']), color, 2)
            cv2.putText(frame,
                        "#" + det_label + ' ' + str(round(obj['confidence'] * 100, 1)) + ' %',
                        (obj['xmin'], obj['ymin'] - 7), cv2.FONT_HERSHEY_COMPLEX, 0.6, color, 1)

        # Draw performance stats over frame
        inf_time_message = "Inference time: N\A for async mode" if is_async_mode else \
            "Inference time: {:.3f} ms".format(det_time * 1e3)
        render_time_message = "OpenCV rendering time: {:.3f} ms".format(render_time * 1e3)
        async_mode_message = "Async mode is on. Processing request {}".format(cur_request_id) if is_async_mode else \
            "Async mode is off. Processing request {}".format(cur_request_id)
        parsing_message = "YOLO parsing time is {:.3f} ms".format(parsing_time * 1e3)

        cv2.putText(frame, inf_time_message, (15, 15), cv2.FONT_HERSHEY_COMPLEX, 0.5, (200, 10, 10), 1)
        cv2.putText(frame, render_time_message, (15, 45), cv2.FONT_HERSHEY_COMPLEX, 0.5, (10, 10, 200), 1)
        cv2.putText(frame, async_mode_message, (10, int(origin_im_size[0] - 20)), cv2.FONT_HERSHEY_COMPLEX, 0.5,
                    (10, 10, 200), 1)
        cv2.putText(frame, parsing_message, (15, 30), cv2.FONT_HERSHEY_COMPLEX, 0.5, (10, 10, 200), 1)

        start_time = time()
        if not args.no_show:
            cv2.imshow("DetectionResults", frame)
        render_time = time() - start_time

        if is_async_mode:
            cur_request_id, next_request_id = next_request_id, cur_request_id
            frame = next_frame

        if not args.no_show:
            key = cv2.waitKey(wait_key_code)
    
            # ESC key
            if key == 27:
                break
            # Tab key
            if key == 9:
                exec_net.requests[cur_request_id].wait()
                is_async_mode = not is_async_mode
                log.info("Switched to {} mode".format("async" if is_async_mode else "sync"))

    cv2.destroyAllWindows()


if __name__ == '__main__':
    sys.exit(main() or 0)

 

Get IR files

  Run following script

python yolov5_demo_OV2021.3.py -i img.jpg -m weight.xml

  Then we can see the inference demo

 

Solutions with some errors

• No module named 'networkx'
  pip install networkx

• ModuleNotFoundError: No module named 'defusedxml'
  pip install defusedxml

• Detected not satisfied dependencies: onnx: installed: 1.8.0, required: >= 1.8.1
  pip install onnx==1.8.1

 

Reference

https://github.com/violet17/yolov5_demo https://github.com/Chen-MingChang/pytorch_YOLO_OpenVINO_demo https://cloud.tencent.com/developer/article/1764073