Rk1126 实现 yolov5 6.2 推理

基于 RK1126 实现 yolov5 6.2 推理.

---

转换 ONNX

1. python export.py --weights ./weights/yolov5s.pt --img 640 --batch 1 --include onnx --simplify

安装 rk 环境

1. 安装部分参考网上, 有很多. 参考: https://github.com/rockchip-linux/rknpu

转换 RK模型 并验证

1. yolov562_to_rknn_3_4.py ( s/m/l/x ..，输出节点不同，使用 netror 查看 )

   # -*- coding: utf-8 -*-
   """
   Created on Wed Oct 12 18:24:38 2022
   
   @author: bobod
   """
   
   
   import os
   import numpy as np
   import cv2
   from rknn.api import RKNN
   
   
   ONNX_MODEL = './weights/yolov5s_v6.2.onnx'
   RKNN_MODEL = './weights/yolov5s_v6.2.rknn'
   IMG_PATH = './000000102411.jpg'
   DATASET = './dataset.txt'
   
   QUANTIZE_ON = True
   
   BOX_THRESH = 0.5
   NMS_THRESH = 0.6
   IMG_SIZE = (640, 640) # (width, height), such as (1280, 736)
   
   SHAPES =((0.0, 0.0), (0.0, 0.0)) #1 scale_coords
   SHAPE =(0,0)
   
   CLASSES = ("person", "bicycle", "car","motorbike ","aeroplane ","bus ","train","truck ","boat","traffic light",
              "fire hydrant","stop sign ","parking meter","bench","bird","cat","dog ","horse ","sheep","cow","elephant",
              "bear","zebra ","giraffe","backpack","umbrella","handbag","tie","suitcase","frisbee","skis","snowboard","sports ball","kite",
              "baseball bat","baseball glove","skateboard","surfboard","tennis racket","bottle","wine glass","cup","fork","knife ",
              "spoon","bowl","banana","apple","sandwich","orange","broccoli","carrot","hot dog","pizza ","donut","cake","chair","sofa",
              "pottedplant","bed","diningtable","toilet ","tvmonitor","laptop	","mouse	","remote ","keyboard ","cell phone","microwave ",
              "oven ","toaster","sink","refrigerator ","book","clock","vase","scissors ","teddy bear ","hair drier", "toothbrush ")
   
   def sigmoid(x):
       return 1 / (1 + np.exp(-x))
   
   def xywh2xyxy(x):
       # Convert [x, y, w, h] to [x1, y1, x2, y2]
       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 process(input, mask, anchors):
   
       anchors = [anchors[i] for i in mask]
       grid_h, grid_w = map(int, input.shape[0:2])
   
       box_confidence = sigmoid(input[..., 4])
       box_confidence = np.expand_dims(box_confidence, axis=-1)
   
       box_class_probs = sigmoid(input[..., 5:])
   
       box_xy = sigmoid(input[..., :2])*2 - 0.5
   
       col = np.tile(np.arange(0, grid_w), grid_h).reshape(-1, grid_w)
       row = np.tile(np.arange(0, grid_h).reshape(-1, 1), grid_w)
       col = col.reshape(grid_h, grid_w, 1, 1).repeat(3, axis=-2)
       row = row.reshape(grid_h, grid_w, 1, 1).repeat(3, axis=-2)
       grid = np.concatenate((col, row), axis=-1)
       box_xy += grid
       box_xy *= (int(IMG_SIZE[1]/grid_h), int(IMG_SIZE[0]/grid_w))
   
       box_wh = pow(sigmoid(input[..., 2:4])*2, 2)
       box_wh = box_wh * anchors
   
       box = np.concatenate((box_xy, box_wh), axis=-1)
   
       return box, box_confidence, box_class_probs
   
   def filter_boxes(boxes, box_confidences, box_class_probs):
       """Filter boxes with box threshold. It's a bit different with origin yolov5 post process!
       # Arguments
           boxes: ndarray, boxes of objects.
           box_confidences: ndarray, confidences of objects.
           box_class_probs: ndarray, class_probs of objects.
       # Returns
           boxes: ndarray, filtered boxes.
           classes: ndarray, classes for boxes.
           scores: ndarray, scores for boxes.
       """
       boxes = boxes.reshape(-1, 4)
       box_confidences = box_confidences.reshape(-1)
       box_class_probs = box_class_probs.reshape(-1, box_class_probs.shape[-1])
   
       _box_pos = np.where(box_confidences >= BOX_THRESH)
       boxes = boxes[_box_pos]
       box_confidences = box_confidences[_box_pos]
       box_class_probs = box_class_probs[_box_pos]
   
       class_max_score = np.max(box_class_probs, axis=-1)
       classes = np.argmax(box_class_probs, axis=-1)
       _class_pos = np.where(class_max_score* box_confidences >= BOX_THRESH)
   
       boxes = boxes[_class_pos]
       classes = classes[_class_pos]
       scores = (class_max_score* box_confidences)[_class_pos]
   
       return boxes, classes, scores
   
   def nms_boxes(boxes, scores):
       """Suppress non-maximal boxes.
       # Arguments
           boxes: ndarray, boxes of objects.
           scores: ndarray, scores of objects.
       # Returns
           keep: ndarray, index of effective boxes.
       """
       x = boxes[:, 0]
       y = boxes[:, 1]
       w = boxes[:, 2] - boxes[:, 0]
       h = boxes[:, 3] - boxes[:, 1]
   
       areas = w * h
       order = scores.argsort()[::-1]
   
       keep = []
       while order.size > 0:
           i = order[0]
           keep.append(i)
   
           xx1 = np.maximum(x[i], x[order[1:]])
           yy1 = np.maximum(y[i], y[order[1:]])
           xx2 = np.minimum(x[i] + w[i], x[order[1:]] + w[order[1:]])
           yy2 = np.minimum(y[i] + h[i], y[order[1:]] + h[order[1:]])
   
           w1 = np.maximum(0.0, xx2 - xx1 + 0.00001)
           h1 = np.maximum(0.0, yy2 - yy1 + 0.00001)
           inter = w1 * h1
   
           ovr = inter / (areas[i] + areas[order[1:]] - inter)
           inds = np.where(ovr <= NMS_THRESH)[0]
           order = order[inds + 1]
       keep = np.array(keep)
       return keep
   
   
   def yolov5_post_process(input_data):
       masks = [[0, 1, 2], [3, 4, 5], [6, 7, 8]]
       anchors = [[10, 13], [16, 30], [33, 23], [30, 61], [62, 45],
                 [59, 119], [116, 90], [156, 198], [373, 326]]
   
       boxes, classes, scores = [], [], []
       for input,mask in zip(input_data, masks):
           b, c, s = process(input, mask, anchors)
           b, c, s = filter_boxes(b, c, s)
           boxes.append(b)
           classes.append(c)
           scores.append(s)
   
       boxes = np.concatenate(boxes)
       boxes = xywh2xyxy(boxes)
       classes = np.concatenate(classes)
       scores = np.concatenate(scores)
   
       nboxes, nclasses, nscores = [], [], []
       for c in set(classes):
           inds = np.where(classes == c)
           b = boxes[inds]
           c = classes[inds]
           s = scores[inds]
   
           keep = nms_boxes(b, s)
   
           nboxes.append(b[keep])
           nclasses.append(c[keep])
           nscores.append(s[keep])
   
       if not nclasses and not nscores:
           return None, None, None
   
       boxes = np.concatenate(nboxes)
       scale_coords(IMG_SIZE, boxes, SHAPE, SHAPES) #2
       classes = np.concatenate(nclasses)
       scores = np.concatenate(nscores)
   
       return boxes, classes, scores
   
   def draw(image, boxes, scores, classes):
       """Draw the boxes on the image.
       # Argument:
           image: original image.
           boxes: ndarray, boxes of objects.
           classes: ndarray, classes of objects.
           scores: ndarray, scores of objects.
           all_classes: all classes name.
       """
       for box, score, cl in zip(boxes, scores, classes):
           left, top, right, bottom = box
           print('class: {}, score: {}'.format(CLASSES[cl], score))
           print('box coordinate left,top,right,bottom: [{}, {}, {}, {}]'.format(left, top, right, bottom))
           left = int(left)
           top = int(top)
           right = int(right)
           bottom = int(bottom)
   
           cv2.rectangle(image, (left, top), (right, bottom), (255, 0, 0), 2)
           cv2.putText(image, '{0} {1:.2f}'.format(CLASSES[cl], score),
                       (left, top - 6),
                       cv2.FONT_HERSHEY_SIMPLEX,
                       0.6, (0, 0, 255), 2)
   
   
   def letterbox(im, new_shape=(640, 640), color=(0, 0, 0)):
       # Resize and pad image while meeting stride-multiple constraints
       shape = im.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])
   
       # 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
   
       dw /= 2  # divide padding into 2 sides
       dh /= 2
   
       if shape[::-1] != new_unpad:  # resize
           im = cv2.resize(im, 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))
       im = cv2.copyMakeBorder(im, top, bottom, left, right, cv2.BORDER_CONSTANT, value=color)  # add border
       return im, ratio, (dw, dh)
   
   #3
   def scale_coords(img1_shape, coords, img0_shape, ratio_pad=None):
       # 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
           pad = (img1_shape[1] - img0_shape[1] * gain) / 2, (img1_shape[0] - img0_shape[0] * gain) / 2  # wh padding
       else:
           gain = ratio_pad[0][0]
           pad = ratio_pad[1]
   
       coords[:, [0, 2]] -= pad[0]  # x padding
       coords[:, [1, 3]] -= pad[1]  # y padding
       coords[:, :4] /= gain
       clip_coords(coords, img0_shape)
       return coords
   
   
   def clip_coords(boxes, shape):
       # Clip bounding xyxy bounding boxes to image shape (height, width)
       boxes[:, [0, 2]] = boxes[:, [0, 2]].clip(0, shape[1])  # x1, x2
       boxes[:, [1, 3]] = boxes[:, [1, 3]].clip(0, shape[0])  # y1, y2
       
   if __name__ == '__main__':
   
       # Create RKNN object
       rknn = RKNN(verbose=False)
   
       if not os.path.exists(ONNX_MODEL):
           print('model not exist')
           exit(-1)
   
       _force_builtin_perm = False
       # pre-process config
       print('--> Config model')
       rknn.config(
                   reorder_channel='0 1 2',
                   mean_values=[[0, 0, 0]],
                   std_values=[[255, 255, 255]],
                   optimization_level=3,
                   #target_platform = 'rk1808',
                   # target_platform='rv1109',
                   target_platform = 'rv1126',
                   quantize_input_node= QUANTIZE_ON,
                   output_optimize=1,
                   force_builtin_perm=_force_builtin_perm)
       print('done')
   
       # Load ONNX model
       print('--> Loading model')
       #ret = rknn.load_pytorch(model=PT_MODEL, input_size_list=[[3,IMG_SIZE[1], IMG_SIZE[0]]])
       ret = rknn.load_onnx(model=ONNX_MODEL, outputs=['output', '391', '402'])
       if ret != 0:
           print('Load yolov5 failed!')
           exit(ret)
       print('done')
   
       # Build model
       print('--> Building model')
       ret = rknn.build(do_quantization=QUANTIZE_ON, dataset=DATASET, pre_compile=False)
       if ret != 0:
           print('Build yolov5 failed!')
           exit(ret)
       print('done')
   
       # Export RKNN model
       print('--> Export RKNN model')
       ret = rknn.export_rknn(RKNN_MODEL)
       if ret != 0:
           print('Export yolov5rknn failed!')
           exit(ret)
       print('done')
   
       # init runtime environment
       print('--> Init runtime environment')
       ret = rknn.init_runtime() 
       #ret = rknn.init_runtime('rv1126', device_id='bab4d7a824f04867')
       # ret = rknn.init_runtime('rv1109', device_id='1109')
       # ret = rknn.init_runtime('rk1808', device_id='1808')
       if ret != 0:
           print('Init runtime environment failed')
           exit(ret)
       print('done')
   
       # Set inputs
       original_img = cv2.imread(IMG_PATH)#4
       img, ratio, pad = letterbox(original_img, new_shape=(IMG_SIZE[1], IMG_SIZE[0]))
       SHAPES=(ratio,pad)
       SHAPE=(original_img.shape[0],original_img.shape[1])
       img = cv2.cvtColor(img, cv2.COLOR_BGR2RGB)
   
       # Inference
       print('--> Running model')
       outputs = rknn.inference(inputs=[img], inputs_pass_through=[0 if not _force_builtin_perm else 1])
   
       # post process
       input0_data = outputs[0]
       input1_data = outputs[1]
       input2_data = outputs[2]
   
       input0_data = input0_data.reshape([3,-1]+list(input0_data.shape[-2:]))
       input1_data = input1_data.reshape([3,-1]+list(input1_data.shape[-2:]))
       input2_data = input2_data.reshape([3,-1]+list(input2_data.shape[-2:]))
   
       input_data = list()
       input_data.append(np.transpose(input0_data, (2, 3, 0, 1)))
       input_data.append(np.transpose(input1_data, (2, 3, 0, 1)))
       input_data.append(np.transpose(input2_data, (2, 3, 0, 1)))
   
   
       boxes, classes, scores = yolov5_post_process(input_data)
   
   
       if boxes is not None:
           draw(original_img, boxes, scores, classes)
       cv2.imwrite("result.jpg", original_img)
   
   

混合量化

1. 首先做精度分析, 量化 build 之后， 调用精度分析函数。后来尝试在 hybrid_quantization_step2 后调用，结果总是不成功 , 原因未知, 有知道的大佬望告知一下.
2. 精度分析: entire_qnt ( 完全量化结果 ) ， fp32 ( fp32结果 ) ，individual_qnt ( 逐层量化结果,即输入为float, 排除累计误差 ), entire_qnt_error_analysis.txt individual_qnt_error_analysis.txt ( 完全量化和逐层量化分析结果 (欧式距离和余弦距离) )
3. 这里需要注意, DATASET 只能有一行数据.

   ...
       # Build model
       print('--> Building model')
       ret = rknn.build(do_quantization=QUANTIZE_ON, dataset=DATASET, pre_compile=False)
       if ret != 0:
           print('Build yolov5 failed!')
           exit(ret)
       print('done')
   
       print('--> Accuracy analysis')
       ret = rknn.accuracy_analysis(inputs=DATASET1,output_dir="./output_dir")
       if ret != 0:
           print('accuracy_analysis failed!')
           exit(ret)
       print('done')
   

4. 生成混合量化配置文件, 调用 rknn.hybrid_quantization_step1 会得到 torchjitexport.data ， torchjitexport.json， torchjitexport.quantization.cfg 3个文件.
5. 根据精度分析结果，torchjitexport.quantization.cfg 将不想量化的层添加到自定义层中.

   # 1.冒号后面加个空格  2.一个模型最多同时只能存在两种量化方式
   # add layer name and corresponding quantized_dtype to customized_quantize_layers, e.g conv2_3: float32
   customized_quantize_layers: {
       "Conv_Conv_0_187": float32,
       "Sigmoid_Sigmoid_1_188_Mul_Mul_2_172": float32,
       "Conv_Conv_3_171": float32,
       ...
   }
   ...
   

6. 调用 rknn.hybrid_quantization_step2 导出rknn模型

   ...
   ret = rknn.hybrid_quantization_step2(model_input='./torchjitexport.json',data_input='./torchjitexport.data', model_quantization_cfg='./torchjitexport.quantization.cfg',dataset=DATASET, pre_compile=False)
       if ret != 0:
           print("hybrid_quantization_step2 failed. ")
           exit(ret)
   
   # Export RKNN model
   print('--> Export RKNN model')
   ret = rknn.export_rknn(RKNN_MODEL)
   if ret != 0:
       print('Export yolov5rknn failed!')
       exit(ret)
   print('done')
   

7. 重复测试模型，找出一个平衡速度和准确率的模型。
8. 验证没问题了，就可以连接板子尝试啦，可以将 pre_compile=True, 提高初始化速度. 不过不能在仿真环境中测试.

---

参考

1. https://github.com/shaoshengsong/rockchip_rknn_yolov5
2. https://github.com/rockchip-linux/rknpu