DongHappyyy

【代码】固定随机数种子，让每次训练的结果保持一致。

Keras/Tensorflow(1.13) 加载模型后占用高显存解决

tensorflow lite 量化

视觉分类任务中可视化工具 t-SNE

Confusion Matrix (混淆矩阵) 的绘制

1、求网络模型参数：n_parameters = sum(p.numel() for p in model.parameters() if p.requires_grad)print('number of params:', n_parameters)2、打印模型参数名称，以及创建参数字典 print(f"\n{'#'*20}\n print the parameters of model\n") for n, p in model_without_ddp.na

reference::对抗神经网络九--反卷积_哔哩哔哩_bilibilihttps://www.bilibili.com/video/BV1y5411V7pU?from=search&seid=931186217418602074&spm_id_from=333.337.0.047.2 转置卷积是一种卷积【动手学深度学习v2】_哔哩哔哩_bilibilihttps://www.bilibili.com/video/BV1CM4y1K7r7?fro...

import sysimport osimport jsonfrom itertools import chainSTART_BOUNDING_BOX_ID = 1 # testing need to change 701# If necessary, pre-define category and its idPRE_DEFINE_CATEGORIES = {"a": 1, "d": 2}def convert(jsonsFile, json_file, img.

多图合成一图：import osimport tqdmimport globimport jsonimport mathimport randomimport numpy as npfrom pathlib import Pathfrom PIL import Image, ImageDrawforesuffix = '_foreground.jpg'pseudosuffix = '_pseudo.jpg'jsonsuffix = '.jpg.json'# 旋转图片，没

import cv2import randomimport json, osfrom pycocotools.coco import COCOfrom skimage import iofrom matplotlib import pyplot as pltcoco_classes = ["111", "112", "113", "121", "122", "123", "131", "132", "133", "211", "212", "213", "2.