一 知識(shí)背景 3D scan&cloud points(點(diǎn)云)patch-based features,fully convolutional network, deep metric learning, sparse tensors,sparse convolutions, hard negeti Read More
1.Overload your functions by having variable number of input and output argumernt.Not only can we overload functions also operators. 我們可以通過(guò)不同的輸入輸出來(lái)重載函 Read More
這篇其實(shí)是TensorFlow的官方tutorials,由于沒(méi)有翻譯,筆者姑且翻譯一下,用來(lái)日后思考。 原址:https://www.tensorflow.org/tutorials/load_data/tf_records 讀取數(shù)據(jù)的效率對(duì)于連續(xù)載入數(shù)據(jù)和將數(shù)據(jù)儲(chǔ)存在文件集(每個(gè)文件大概100-2 Read More