Implementation of yolov5 single image detection based on onnxruntime

Connected to a be based on pytorch Of YOLOv5 Single picture detection implementation , We did pytorch Forward reasoning , But this reasoning process needs to rely on yolov5 Its own model file and the process of structure construction , So it's still troublesome . here , Is there a direct forward reasoning , Then only process the results , Don't need to consider yolov5 Own documents . So now the introduction is based on onnx The reasoning of . The reasoning process is also very simple , Convert the original model into onnx Format , And then use onnxruntime Just do it , See my article for specific operation .

One 、pt turn onnx

Here we mainly refer to ：https://github.com/ultralytics/yolov5/issues/251 Content in , Get into yolov5 The installation directory , Perform the following ：

python models/export.py --weights yolov5s.pt --img 640 --batch 1  

Two 、onnxruntime Forward reasoning

1. Installation dependency

pip install onnxruntime

2. Code implementation

# coding=utf-8
import cv2.cv2 as cv2
import numpy as np
import onnxruntime
import torch
import torchvision
import time
import random

class YOLOV5_ONNX(object):
    def __init__(self,onnx_path):
        ''' initialization onnx'''
        self.onnx_session=onnxruntime.InferenceSession(onnx_path)
        self.input_name=self.get_input_name()
        self.output_name=self.get_output_name()

    def get_input_name(self):
        ''' Get input node name '''
        input_name=[]
        for node in self.onnx_session.get_inputs():
            input_name.append(node.name)

        return input_name


    def get_output_name(self):
        ''' Get the output node name '''
        output_name=[]
        for node in self.onnx_session.get_outputs():
            output_name.append(node.name)

        return output_name

    def get_input_feed(self,image_tensor):
        ''' Get input tensor'''
        input_feed={
    }
        for name in self.input_name:
            input_feed[name]=image_tensor

        return input_feed

    def letterbox(self,img, new_shape=(640, 640), color=(114, 114, 114), auto=False, scaleFill=False, scaleup=True,
                  stride=32):
        ''' Picture normalization '''
        # Resize and pad image while meeting stride-multiple constraints
        shape = img.shape[:2]  # current shape [height, width]
        if isinstance(new_shape, int):
            new_shape = (new_shape, new_shape)

        # Scale ratio (new / old)
        r = min(new_shape[0] / shape[0], new_shape[1] / 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 = new_shape[1] - new_unpad[0], new_shape[0] - new_unpad[1]  # wh padding

        if auto:  # minimum rectangle
            dw, dh = np.mod(dw, stride), np.mod(dh, stride)  # wh padding
        elif scaleFill:  # stretch
            dw, dh = 0.0, 0.0
            new_unpad = (new_shape[1], new_shape[0])
            ratio = new_shape[1] / shape[1], new_shape[0] / 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
        return img, ratio, (dw, dh)

    def xywh2xyxy(self,x):
        # Convert nx4 boxes from [x, y, w, h] to [x1, y1, x2, y2] where xy1=top-left, xy2=bottom-right
        y = np.copy(x)

        y[:, 0] = x[:, 0] - x[:, 2] / 2  # top left x
        y[:, 1] = x[:, 1] - x[:, 3] / 2  # top left y
        y[:, 2] = x[:, 0] + x[:, 2] / 2  # bottom right x
        y[:, 3] = x[:, 1] + x[:, 3] / 2  # bottom right y

        return y

    def nms(self,prediction, conf_thres=0.1, iou_thres=0.6, agnostic=False):
        if prediction.dtype is torch.float16:
            prediction = prediction.float()  # to FP32
        xc = prediction[..., 4] > conf_thres  # candidates
        min_wh, max_wh = 2, 4096  # (pixels) minimum and maximum box width and height
        max_det = 300  # maximum number of detections per image
        output = [None] * prediction.shape[0]
        for xi, x in enumerate(prediction):  # image index, image inference
            x = x[xc[xi]]  # confidence
            if not x.shape[0]:
                continue

            x[:, 5:] *= x[:, 4:5]  # conf = obj_conf * cls_conf
            box = self.xywh2xyxy(x[:, :4])

            conf, j = x[:, 5:].max(1, keepdim=True)
            x = torch.cat((torch.tensor(box), conf, j.float()), 1)[conf.view(-1) > conf_thres]
            n = x.shape[0]  # number of boxes
            if not n:
                continue
            c = x[:, 5:6] * (0 if agnostic else max_wh)  # classes
            boxes, scores = x[:, :4] + c, x[:, 4]  # boxes (offset by class), scores
            i = torchvision.ops.boxes.nms(boxes, scores, iou_thres)
            if i.shape[0] > max_det:  # limit detections
                i = i[:max_det]
            output[xi] = x[i]

        return output

    def clip_coords(self,boxes, img_shape):
        ''' Check to see if it's out of bounds '''
        # Clip bounding xyxy bounding boxes to image shape (height, width)
        boxes[:, 0].clamp_(0, img_shape[1])  # x1
        boxes[:, 1].clamp_(0, img_shape[0])  # y1
        boxes[:, 2].clamp_(0, img_shape[1])  # x2
        boxes[:, 3].clamp_(0, img_shape[0])  # y2

    def scale_coords(self,img1_shape, coords, img0_shape, ratio_pad=None):
        '''  The coordinates correspond to the original image , Reverse operation ： subtract pad, Divided by the minimum zoom factor  :param img1_shape:  Enter dimensions  :param coords:  Enter coordinates  :param img0_shape:  Mapped dimensions  :param ratio_pad: :return: '''

        # Rescale coords (xyxy) from img1_shape to img0_shape
        if ratio_pad is None:  # calculate from img0_shape
            gain = min(img1_shape[0] / img0_shape[0], img1_shape[1] / img0_shape[1])  # gain = old / new, Calculate the zoom ratio 
            pad = (img1_shape[1] - img0_shape[1] * gain) / 2, (
                        img1_shape[0] - img0_shape[0] * gain) / 2  # wh padding , Calculate the expanded size 
        else:
            gain = ratio_pad[0][0]
            pad = ratio_pad[1]

        coords[:, [0, 2]] -= pad[0]  # x padding, subtract x Expansion in direction 
        coords[:, [1, 3]] -= pad[1]  # y padding, subtract y Expansion in direction 
        coords[:, :4] /= gain  #  take box The coordinates correspond to the original image 
        self.clip_coords(coords, img0_shape)  #  Boundary check 
        return coords

    def sigmoid(self,x):
        return 1 / (1 + np.exp(-x))



    def infer(self,img_path):
        ''' Perform the forward operation to predict the output '''
        #  Super parameter settings 
        img_size=(640,640) # Picture zoom size 
        conf_thres=0.25 # Confidence threshold 
        iou_thres=0.45 #iou threshold 
        class_num=1 # Number of categories 

        stride=[8,16,32]

        anchor_list= [[10,13, 16,30, 33,23],[30,61, 62,45, 59,119], [116,90, 156,198, 373,326]]
        anchor = np.array(anchor_list).astype(np.float).reshape(3,-1,2)

        area = img_size[0] * img_size[1]
        size = [int(area / stride[0] ** 2), int(area / stride[1] ** 2), int(area / stride[2] ** 2)]
        feature = [[int(j / stride[i]) for j in img_size] for i in range(3)]


        #  Read the picture 
        src_img=cv2.imread(img_path)
        src_size=src_img.shape[:2]

        #  Picture filling and normalization 
        img=self.letterbox(src_img,img_size,stride=32)[0]

        # Convert
        img = img[:, :, ::-1].transpose(2, 0, 1)  # BGR to RGB, to 3x416x416
        img = np.ascontiguousarray(img)


        #  normalization 
        img=img.astype(dtype=np.float32)
        img/=255.0

        # # BGR to RGB
        # img = img[:, :, ::-1].transpose(2, 0, 1)
        # img = np.ascontiguousarray(img)

        #  Dimension expansion 
        img=np.expand_dims(img,axis=0)

        #  Forward reasoning 
        start=time.time()
        input_feed=self.get_input_feed(img)
        pred=self.onnx_session.run(output_names=self.output_name,input_feed=input_feed)

        # Extract features 
        y = []
        y.append(torch.tensor(pred[0].reshape(-1,size[0]*3,5+class_num)).sigmoid())
        y.append(torch.tensor(pred[1].reshape(-1,size[1]*3,5+class_num)).sigmoid())
        y.append(torch.tensor(pred[2].reshape(-1,size[2]*3,5+class_num)).sigmoid())

        grid = []
        for k, f in enumerate(feature):
            grid.append([[i, j] for j in range(f[0]) for i in range(f[1])])

        z = []
        for i in range(3):
            src = y[i]

            xy = src[..., 0:2] * 2. - 0.5
            wh = (src[..., 2:4] * 2) ** 2
            dst_xy = []
            dst_wh = []
            for j in range(3):
                dst_xy.append((xy[:, j * size[i]:(j + 1) * size[i], :] + torch.tensor(grid[i])) * stride[i])
                dst_wh.append(wh[:, j * size[i]:(j + 1) * size[i], :] * anchor[i][j])
            src[..., 0:2] = torch.from_numpy(np.concatenate((dst_xy[0], dst_xy[1], dst_xy[2]), axis=1))
            src[..., 2:4] = torch.from_numpy(np.concatenate((dst_wh[0], dst_wh[1], dst_wh[2]), axis=1))
            z.append(src.view(1, -1, 5+class_num))

        results = torch.cat(z, 1)
        results = self.nms(results, conf_thres, iou_thres)
        cast=time.time()-start
        print("cast time:{}".format(cast))



        # Map to the original image 
        img_shape=img.shape[2:]
        print(img_size)
        for det in results:  # detections per image
            if det is not None and len(det):
                det[:, :4] = self.scale_coords(img_shape, det[:, :4],src_size).round()

        if det is not None and len(det):
            self.draw(src_img, det)




    def plot_one_box(self,x, img, color=None, label=None, line_thickness=None):
        # Plots one bounding box on image img
        tl = line_thickness or round(0.002 * (img.shape[0] + img.shape[1]) / 2) + 1  # line/font thickness
        color = color or [random.randint(0, 255) for _ in range(3)]
        c1, c2 = (int(x[0]), int(x[1])), (int(x[2]), int(x[3]))
        cv2.rectangle(img, c1, c2, color, thickness=tl, lineType=cv2.LINE_AA)
        if label:
            tf = max(tl - 1, 1)  # font thickness
            t_size = cv2.getTextSize(label, 0, fontScale=tl / 3, thickness=tf)[0]
            c2 = c1[0] + t_size[0], c1[1] - t_size[1] - 3
            cv2.rectangle(img, c1, c2, color, -1, cv2.LINE_AA)  # filled
            cv2.putText(img, label, (c1[0], c1[1] - 2), 0, tl / 3, [225, 255, 255], thickness=tf, lineType=cv2.LINE_AA)

    def draw(self,img, boxinfo):
        colors = [[0, 0, 255]]
        for *xyxy, conf, cls in boxinfo:
            label = '%s %.2f' % ('image', conf)
            print('xyxy: ', xyxy)
            self.plot_one_box(xyxy, img, label=label, color=colors[int(cls)], line_thickness=1)

        cv2.namedWindow("dst",0)
        cv2.imshow("dst", img)
        cv2.imwrite("data/res1.jpg",img)
        cv2.waitKey(0)
        # cv2.imencode('.jpg', img)[1].tofile(os.path.join(dst, id + ".jpg"))
        return 0


if __name__=="__main__":
    model=YOLOV5_ONNX(onnx_path="./weights/image_detect.onnx")
    model.infer(img_path="data/PMC2663376_00004.jpg")

result ：

3、onnxruntime and pytorch Compare

• onnxruntime Reasoning time
  
• pytorch Reasoning time
  

We are normalizing to 640x640 Compare on the image ,onnx Inferential than pure pytorch Time has improved 1 times . explain onnx Reasoning is still possible , Later, it will be tested on other acceleration frameworks , Look forward to the follow-up ...

github link ：yolov5 Forward reasoning implementation

Reference link ：
onnxruntime-for-yolov5
python3 onnx Reasoning Demo