1. Yolov5簡(jiǎn)介
YOLOv5 模型是 Ultralytics 公司于 2020 年 6 月 9 日公開(kāi)發(fā)布的。YOLOv5 模型是基于 YOLOv3 模型基礎(chǔ)上改進(jìn)而來(lái)的,有 YOLOv5s、YOLOv5m、YOLOv5l、YOLOv5x 四個(gè)模型。YOLOv5 相比YOLOv4 而言,在檢測(cè)平均精度降低不多的基礎(chǔ)上,具有均值權(quán)重文件更小,訓(xùn)練時(shí)間和推理速度更短的特點(diǎn)。YOLOv5 的網(wǎng)絡(luò)結(jié)構(gòu)分為輸入端BackboneNeck、Head 四個(gè)部分。本教程針對(duì)目標(biāo)檢測(cè)算法yolov5的訓(xùn)練和部署到EASY-EAI-Nano(RV1126)進(jìn)行說(shuō)明,而數(shù)據(jù)標(biāo)注方法可以參考我們往期的文章《Labelimg的安裝與使用》。以下為YOLOv5訓(xùn)練部署的大致流程:
2. 準(zhǔn)備數(shù)據(jù)集
2.1 數(shù)據(jù)集下載
本教程以口罩檢測(cè)為例,數(shù)據(jù)集的百度網(wǎng)盤(pán)下載鏈接為:
https://pan.baidu.com/s/1vtxWurn1Mqu-wJ017eaQrw 提取碼:6666
解壓完成后得到以下三個(gè)文件:
2.2 生成路徑列表
在數(shù)據(jù)集目錄下執(zhí)行腳本list_dataset_file.py:
python list_dataset_file.py
執(zhí)行現(xiàn)象如下圖所示:
得到訓(xùn)練樣本列表文件train.txt和驗(yàn)證樣本列表文件valid.txt,如下圖所示:
3. 下載yolov5訓(xùn)練源碼
通過(guò)git工具,在PC端克隆遠(yuǎn)程倉(cāng)庫(kù)(注:此處可能會(huì)因網(wǎng)絡(luò)原因造成卡頓,請(qǐng)耐心等待):
git clone https://github.com/EASY-EAI/yolov5.git
得到下圖所示目錄:
4. 訓(xùn)練算法模型
切換到y(tǒng)olov5的工作目錄,接下來(lái)以訓(xùn)練一個(gè)口罩檢測(cè)模型為例進(jìn)行說(shuō)明。需要修改data/mask.yaml里面的train.txt和valid.txt的路徑。
執(zhí)行下列腳本訓(xùn)練算法模型:
python train.py --data mask.yaml --cfg yolov5s.yaml --weights "" --batch-size 64
開(kāi)始訓(xùn)練模型,如下圖所示:
關(guān)于算法精度結(jié)果可以查看./runs/train/results.csv獲得。
5. 在PC端進(jìn)行模型預(yù)測(cè)
訓(xùn)練完畢后,在./runs/train/exp/weights/best.pt生成通過(guò)驗(yàn)證集測(cè)試的最好結(jié)果的模型。同時(shí)可以執(zhí)行模型預(yù)測(cè),初步評(píng)估模型的效果:
python detect.py --source data/images --weights ./runs/train/exp/weights/best.pt --conf 0.5
6. pt模型轉(zhuǎn)換為onnx模型
算法部署到EASY-EAI-Nano需要轉(zhuǎn)換為RKNN模型,而轉(zhuǎn)換RKNN之前可以把模型先轉(zhuǎn)換為ONNX模型,同時(shí)會(huì)生成best.anchors.txt:
python export.py --include onnx --rknpu RV1126 --weights ./runs/train/exp/weights/best.pt
生成如下圖所示:
7. 轉(zhuǎn)換為rknn模型環(huán)境搭建
onnx模型需要轉(zhuǎn)換為rknn模型才能在EASY-EAI-Nano運(yùn)行,所以需要先搭建rknn-toolkit模型轉(zhuǎn)換工具的環(huán)境。當(dāng)然tensorflow、tensroflow lite、caffe、darknet等也是通過(guò)類(lèi)似的方法進(jìn)行模型轉(zhuǎn)換,只是本教程onnx為例。
7.1 概述
模型轉(zhuǎn)換環(huán)境搭建流程如下所示:
7.2 下載模型轉(zhuǎn)換工具
為了保證模型轉(zhuǎn)換工具順利運(yùn)行,請(qǐng)下載網(wǎng)盤(pán)里”AI算法開(kāi)發(fā)/RKNN-Toolkit模型轉(zhuǎn)換工具/rknn-toolkit-v1.7.1/docker/rknn-toolkit-1.7.1-docker.tar.gz”。網(wǎng)盤(pán)下載鏈接:https://pan.baidu.com/s/1LUtU_-on7UB3kvloJlAMkA 提取碼:teuc
7.3 把工具移到ubuntu18.04
把下載完成的docker鏡像移到我司的虛擬機(jī)ubuntu18.04的rknn-toolkit目錄,如下圖所示:
7.4 運(yùn)行模型轉(zhuǎn)換工具環(huán)境
7.4.1 打開(kāi)終端
在該目錄打開(kāi)終端:
7.4.2 加載docker鏡像
執(zhí)行以下指令加載模型轉(zhuǎn)換工具docker鏡像:
docker load --input /home/developer/rknn-toolkit/rknn-toolkit-1.7.1-docker.tar.gz
7.4.3 進(jìn)入鏡像bash環(huán)境
執(zhí)行以下指令進(jìn)入鏡像bash環(huán)境:
docker run -t -i --privileged -v /dev/bus/usb:/dev/bus/usb rknn-toolkit:1.7.1 /bin/bash
現(xiàn)象如下圖所示:
7.4.4 測(cè)試環(huán)境
輸入“python”加載python相關(guān)庫(kù),嘗試加載rknn庫(kù),如下圖環(huán)境測(cè)試成功:
至此,模型轉(zhuǎn)換工具環(huán)境搭建完成。
8. rknn模型轉(zhuǎn)換流程介紹
EASY EAI Nano支持.rknn后綴的模型的評(píng)估及運(yùn)行,對(duì)于常見(jiàn)的tensorflow、tensroflow lite、caffe、darknet、onnx和Pytorch模型都可以通過(guò)我們提供的 toolkit 工具將其轉(zhuǎn)換至 rknn 模型,而對(duì)于其他框架訓(xùn)練出來(lái)的模型,也可以先將其轉(zhuǎn)至 onnx 模型再轉(zhuǎn)換為 rknn 模型。模型轉(zhuǎn)換操作流程如下圖所示:
8.1 模型轉(zhuǎn)換Demo下載
下載百度網(wǎng)盤(pán)鏈接:https://pan.baidu.com/s/1uAiQ6edeGIDvQ7HAm7p0jg
提取碼:6666把model_convert.tar.bz2解壓到虛擬機(jī),如下圖所示:
8.2 進(jìn)入模型轉(zhuǎn)換工具docker環(huán)境
執(zhí)行以下指令把工作區(qū)域映射進(jìn)docker鏡像,其中/home/developer/rknn-toolkit/model_convert為工作區(qū)域,/test為映射到docker鏡像,/dev/bus/usb:/dev/bus/usb為映射usb到docker鏡像:
docker run -t -i --privileged -v /dev/bus/usb:/dev/bus/usb -v /home/developer/rknn-toolkit/model_convert:/test rknn-toolkit:1.7.1 /bin/bash
執(zhí)行成功如下圖所示:
8.3 模型轉(zhuǎn)換操作說(shuō)明
8.3.1 模型轉(zhuǎn)換Demo目錄結(jié)構(gòu)
模型轉(zhuǎn)換測(cè)試Demo由mask_object_detect和quant_dataset組成。coco_object_detect存放軟件腳本,quant_dataset存放量化模型所需的數(shù)據(jù)。如下圖所示:
mask_object_detect文件夾存放以下內(nèi)容,如下圖所示:
8.3.2 生成量化圖片列表
在docker環(huán)境切換到模型轉(zhuǎn)換工作目錄:
cd /test/mask_object_detect/
如下圖所示:
執(zhí)行g(shù)en_list.py生成量化圖片列表:
python gen_list.py
命令行現(xiàn)象如下圖所示:
生成“量化圖片列表”如下文件夾所示:
8.3.3 onnx模型轉(zhuǎn)換為rknn模型
rknn_convert.py腳本默認(rèn)進(jìn)行int8量化操作,腳本代碼清單如下所示:
import os import urllib import traceback import time import sys import numpy as np import cv2 from rknn.api import RKNN ONNX_MODEL = 'best.onnx' RKNN_MODEL = './yolov5_mask_rv1126.rknn' DATASET = './pic_path.txt' QUANTIZE_ON = True if __name__ == '__main__': # Create RKNN object rknn = RKNN(verbose=True) if not os.path.exists(ONNX_MODEL): print('model not exist') exit(-1) # 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 = 'rv1126', output_optimize=1, quantize_input_node=QUANTIZE_ON) print('done') # Load ONNX model print('--> Loading model') ret = rknn.load_onnx(model=ONNX_MODEL) 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) 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')
把onnx模型best.onnx放到mask_object_detect目錄,并執(zhí)行rknn_convert.py腳本進(jìn)行模型轉(zhuǎn)換:
python rknn_convert.py
生成模型如下圖所示,此模型可以在rknn環(huán)境和EASY EAI Nano環(huán)境運(yùn)行:
8.3.4 運(yùn)行rknn模型
用yolov5_mask_test.py腳本在PC端的環(huán)境下可以運(yùn)行rknn的模型,如下圖所示:
yolov5_mask_test.py腳本程序清單如下所示:
import os import urllib import traceback import time import sys import numpy as np import cv2 import random from rknn.api import RKNN RKNN_MODEL = 'yolov5_mask_rv1126.rknn' IMG_PATH = './test.jpg' DATASET = './dataset.txt' BOX_THRESH = 0.25 NMS_THRESH = 0.6 IMG_SIZE = 640 CLASSES = ("head", "mask") 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_w).reshape(-1, grid_w) row = np.tile(np.arange(0, grid_h).reshape(-1, 1), grid_h) 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/grid_h) 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. """ box_scores = box_confidences * box_class_probs box_classes = np.argmax(box_class_probs, axis=-1) box_class_scores = np.max(box_scores, axis=-1) pos = np.where(box_confidences[...,0] >= BOX_THRESH) boxes = boxes[pos] classes = box_classes[pos] scores = box_class_scores[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) classes = np.concatenate(nclasses) scores = np.concatenate(nscores) return boxes, classes, scores def scale_coords(x1, y1, x2, y2, dst_width, dst_height): dst_top, dst_left, dst_right, dst_bottom = 0, 0, 0, 0 gain = 0 if dst_width > dst_height: image_max_len = dst_width gain = IMG_SIZE / image_max_len resized_height = dst_height * gain height_pading = (IMG_SIZE - resized_height)/2 print("height_pading:", height_pading) y1 = (y1 - height_pading) y2 = (y2 - height_pading) print("gain:", gain) dst_x1 = int(x1 / gain) dst_y1 = int(y1 / gain) dst_x2 = int(x2 / gain) dst_y2 = int(y2 / gain) return dst_x1, dst_y1, dst_x2, dst_y2 def plot_one_box(x, img, color=None, label=None, line_thickness=None): 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(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): x1, y1, x2, y2 = box print('class: {}, score: {}'.format(CLASSES[cl], score)) print('box coordinate x1,y1,x2,y2: [{}, {}, {}, {}]'.format(x1, y1, x2, y2)) x1 = int(x1) y1 = int(y1) x2 = int(x2) y2 = int(y2) dst_x1, dst_y1, dst_x2, dst_y2 = scale_coords(x1, y1, x2, y2, image.shape[1], image.shape[0]) #print("img.cols:", image.cols) plot_one_box((dst_x1, dst_y1, dst_x2, dst_y2), image, label='{0} {1:.2f}'.format(CLASSES[cl], score)) ''' cv2.rectangle(image, (dst_x1, dst_y1), (dst_x2, dst_y2), (255, 0, 0), 2) cv2.putText(image, '{0} {1:.2f}'.format(CLASSES[cl], score), (dst_x1, dst_y1 - 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) if __name__ == '__main__': # Create RKNN object rknn = RKNN(verbose=True) print('--> Loading model') ret = rknn.load_rknn(RKNN_MODEL) if ret != 0: print('load rknn model failed') exit(ret) print('done') # init runtime environment print('--> Init runtime environment') ret = rknn.init_runtime() # ret = rknn.init_runtime('rv1126', device_id='1126') if ret != 0: print('Init runtime environment failed') exit(ret) print('done') # Set inputs img = cv2.imread(IMG_PATH) letter_img, ratio, (dw, dh) = letterbox(img, new_shape=(IMG_SIZE, IMG_SIZE)) letter_img = cv2.cvtColor(letter_img, cv2.COLOR_BGR2RGB) # Inference print('--> Running model') outputs = rknn.inference(inputs=[letter_img]) print('--> inference done') # 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))) print('--> transpose done') boxes, classes, scores = yolov5_post_process(input_data) print('--> get result done') #img_1 = cv2.cvtColor(img, cv2.COLOR_RGB2BGR) if boxes is not None: draw(img, boxes, scores, classes) cv2.imwrite('./result.jpg', img) #cv2.imshow("post process result", img_1) #cv2.waitKeyEx(0) rknn.release()
執(zhí)行后得到result.jpg如下圖所示:
8.3.5 模型預(yù)編譯
由于rknn模型用NPU API在EASY EAI Nano加載的時(shí)候啟動(dòng)速度會(huì)好慢,在評(píng)估完模型精度沒(méi)問(wèn)題的情況下,建議進(jìn)行模型預(yù)編譯。預(yù)編譯的時(shí)候需要通過(guò)EASY EAI Nano主板的環(huán)境,所以請(qǐng)務(wù)必接上adb口與ubuntu保證穩(wěn)定連接。板子端接線如下圖所示,撥碼開(kāi)關(guān)需要是adb:
虛擬機(jī)要保證接上adb設(shè)備:
由于在虛擬機(jī)里ubuntu環(huán)境與docker環(huán)境對(duì)adb設(shè)備資源是競(jìng)爭(zhēng)關(guān)系,所以需要關(guān)掉ubuntu環(huán)境下的adb服務(wù),且在docker里面通過(guò)apt-get安裝adb軟件包。以下指令在ubuntu環(huán)境與docker環(huán)境里各自執(zhí)行:
在docker環(huán)境里執(zhí)行adb devices,現(xiàn)象如下圖所示則設(shè)備連接成功:
運(yùn)行precompile_rknn.py腳本把模型執(zhí)行預(yù)編譯:
python precompile_rknn.py
執(zhí)行效果如下圖所示,生成預(yù)編譯模型yolov5_mask_rv1126_pre.rknn:
至此預(yù)編譯部署完成,模型轉(zhuǎn)換步驟已全部完成。生成如下預(yù)編譯后的int8量化模型:
9. 模型部署示例
9.1 模型部署示例介紹
本小節(jié)展示yolov5模型的在EASY EAI Nano的部署過(guò)程,該模型僅經(jīng)過(guò)簡(jiǎn)單訓(xùn)練供示例使用,不保證模型精度。
9.2 準(zhǔn)備工作
9.2.1 硬件準(zhǔn)備
EASY EAI Nano開(kāi)發(fā)板,microUSB數(shù)據(jù)線,帶linux操作系統(tǒng)的電腦。需保證EASY EAI Nano與linux系統(tǒng)保持adb連接。
9.2.2 交叉編譯環(huán)境準(zhǔn)備
本示例需要交叉編譯環(huán)境的支持,可以參考在線文檔“入門(mén)指南/開(kāi)發(fā)環(huán)境準(zhǔn)備/安裝交叉編譯工具鏈”。鏈接為:https://www.easy-eai.com/document_details/3/135。
9.2.3 文件下載
下載yolov5 C Demo示例文件。百度網(wǎng)盤(pán)鏈接:https://pan.baidu.com/s/1XmxU9Putp_qSYTSQPqxMDQ提取碼:6666下載解壓后如下圖所示:
9.3 在EASY EAI Nano運(yùn)行yolov5 demo
9.3.1 解壓yolov5 demo
下載程序包移至ubuntu環(huán)境后,執(zhí)行以下指令解壓:
tar -xvf yolov5_detect_C_demo.tar.bz2
9.3.2 編譯yolov5 demo
執(zhí)行以下腳本編譯demo:
./build.sh
編譯成功后如下圖所示:
9.3.3 執(zhí)行yolov5 demo
執(zhí)行以下指令把可執(zhí)行程序推送到開(kāi)發(fā)板端:
adb push yolov5_detect_demo_release/ /userdata
登錄到開(kāi)發(fā)板執(zhí)行程序:
adb shell
執(zhí)行結(jié)果如下圖所示,算法執(zhí)行時(shí)間為50ms:
取回測(cè)試圖片:
adb pull /userdata/yolov5_detect_demo_release/result.jpg .
測(cè)試結(jié)果如下圖所示:
10. 基于攝像頭的AI Demo
10.1 攝像頭Demo介紹
本小節(jié)展示yolov5模型的在EASY EAI Nano執(zhí)行攝像頭Demo的過(guò)程,該模型僅經(jīng)過(guò)簡(jiǎn)單訓(xùn)練供示例使用,不保證模型精度。
10.2 準(zhǔn)備工作
10.2.1 硬件準(zhǔn)備
EASY-EAI-Nano人工智能開(kāi)發(fā)套件(包括:EASY EAI Nano開(kāi)發(fā)板,雙目攝像頭,5寸高清屏幕,microUSB數(shù)據(jù)線),帶linux操作系統(tǒng)的電腦,。需保證EASY EAI Nano與linux系統(tǒng)保持adb連接。
10.2.2 交叉編譯環(huán)境準(zhǔn)備
本示例需要交叉編譯環(huán)境的支持,可以參考在線文檔“入門(mén)指南/開(kāi)發(fā)環(huán)境準(zhǔn)備/安裝交叉編譯工具鏈”。鏈接為:https://www.easy-eai.com/document_details/3/135。
10.2.3 文件下載
攝像頭識(shí)別Demo的程序源碼可以通過(guò)百度網(wǎng)盤(pán)下載:
https://pan.baidu.com/s/18cAp4yT_LhDZ5XAHG-L1lw(提取碼:6666 )。下載解壓后如下圖所示:
10.3 在EASY EAI Nano運(yùn)行yolov5 demo
10.3.1 解壓yolov5 camera demo
下載程序包移至ubuntu環(huán)境后,執(zhí)行以下指令解壓:
tar -xvf yolov5_detect_camera_demo.tar.tar.bz2
10.3.2 編譯yolov5 camera demo
執(zhí)行以下腳本編譯demo:
./build.sh
編譯成功后如下圖所示:
10.3.3 執(zhí)行yolov5 camera demo
執(zhí)行以下指令把可執(zhí)行程序推送到開(kāi)發(fā)板端:
adb push yolov5_detect_camera_demo_release/ /userdata
登錄到開(kāi)發(fā)板執(zhí)行程序:
adb shell
測(cè)試結(jié)果如下圖所示:
11. 資料下載
資料名稱 | 鏈接 |
訓(xùn)練代碼github | https://github.com/EASY-EAI/yolov5 |
算法教程完整源碼包 |
https://pan.baidu.com/s/1-78z8joPYOaGEVFg0I_WZA 提取碼:6666 |
硬件外設(shè)庫(kù)源碼github | https://github.com/EASY-EAI/EASY-EAI-Toolkit-C-SDK |
12. 硬件使用
本教程使用的是EASY EAI nano(RV1126)開(kāi)發(fā)板EASY EAI Nano是基于瑞芯微RV1126 處理器設(shè)計(jì),具有四核CPU@1.5GHz與NPU@2Tops AI邊緣計(jì)算能力。實(shí)現(xiàn)AI運(yùn)算的功耗不及所需GPU的10%。配套AI算法工具完善,支持Tensorflow、Pytorch、Caffe、MxNet、DarkNet、ONNX等主流AI框架直接轉(zhuǎn)換和部署。有豐富的軟硬件開(kāi)發(fā)資料,而且外設(shè)資源豐富,接口齊全,還有豐富的功能配件可供選擇。集成有以太網(wǎng)、Wi-Fi 等通信外設(shè)。攝像頭、顯示屏(帶電容觸摸)、喇叭、麥克風(fēng)等交互外設(shè)。2 路 USB Host 接口、1 路 USB Device 調(diào)試接口。集成協(xié)議串口、TF 卡、IO 拓展接口(兼容樹(shù)莓派/Jetson nano拓展接口)等通用外設(shè)。內(nèi)置人臉識(shí)別、安全帽監(jiān)測(cè)、人體骨骼點(diǎn)識(shí)別、火焰檢測(cè)、車(chē)輛檢測(cè)等各類(lèi) AI 算法,并提供完整的 Linux 開(kāi)發(fā)包供客戶二次開(kāi)發(fā)。
EASY-EAI-Nano產(chǎn)品在線文檔:
https://www.easy-eai.com/document_details/3/143
本篇文章到這里就結(jié)束啦,關(guān)于我們更多請(qǐng)前往官網(wǎng)了解
https://www.easy-eai.com
后續(xù)更多教程關(guān)注我們,不定時(shí)更新各種新品及活動(dòng)
-
算法
+關(guān)注
關(guān)注
23文章
4552瀏覽量
92020 -
開(kāi)發(fā)板
+關(guān)注
關(guān)注
25文章
4771瀏覽量
96162 -
模型
+關(guān)注
關(guān)注
1文章
3032瀏覽量
48356 -
easy
+關(guān)注
關(guān)注
0文章
18瀏覽量
12521 -
Bash
+關(guān)注
關(guān)注
0文章
57瀏覽量
10134
發(fā)布評(píng)論請(qǐng)先 登錄
相關(guān)推薦
評(píng)論