静谧、淡雅

目标检测领域数据集VOC+COCO数据集下载地址:https://zhuanlan.zhihu.com/p/340900096

1 检测框可视化代码：import cv2from pycocotools.coco import COCOdef displayInfo(dataset_dir, json_file): coco = COCO(json_file) catIds = coco.getCatIds(catNms=['donut']) imgIds = coco.getImgIds(catIds=catIds) for i in range(len(imgIds)):

1 coco数据集json转成基于每个图像的from __future__ import print_functionimport skimage.io as ioimport jsonfrom pycocotools.coco import COCOjson_file = 'C:/Users/DELL/Desktop/annotations/instances_val2017.json'data = json.load(open(json_file, 'r'))data_temp = {}

COCO：https://cocodataset.org/#detections-eval上图中出现-1的情况，初始值为-1。模型检测不出中小目标或者没有中小目标，都可能出现上述情况。????