Python
Python
coolsunxu
仅仅记录下自己的成长,不喜勿喷
展开
-
视频动作标注器
视频动作标注器原创 2021-02-22 07:35:58 · 297 阅读 · 0 评论 -
BiTraP:Bi-directional Pedestrian Trajectory Prediction with Multi-modal Goal Estimation
BiTraP:Bi-directional Pedestrian Trajectory Prediction with Multi-modal Goal Estimation原创 2020-08-05 15:42:10 · 648 阅读 · 1 评论 -
python 图像重叠分块恢复
python 图像重叠分块恢复原创 2020-07-24 09:05:14 · 1208 阅读 · 3 评论 -
Windows10 tensorflow2.2 ImportError: DLL load failed 问题
Windows10 tensorflow2.2 ImportError: DLL load failed 问题原创 2020-06-13 22:31:49 · 395 阅读 · 0 评论 -
ssd pytorch eval 报错解决办法
ssd pytorch eval 报错解决办法原创 2020-06-11 11:29:34 · 1177 阅读 · 0 评论 -
CVAE(tensorflow)
CVAE(tensorflow)原创 2020-05-24 21:42:49 · 664 阅读 · 0 评论 -
tf.keras.layers.Conv2DTranspose中反卷积输入和输出关系
tf.keras.layers.Conv2DTranspose中反卷积输入和输出关系原创 2020-05-24 16:59:39 · 3912 阅读 · 0 评论 -
GeeksforGeeks(AVL)
GeeksforGeeks(AVL)原创 2020-05-23 09:58:08 · 243 阅读 · 0 评论 -
GraphTCN: Spatio-Temporal Interaction Modeling for Human Trajectory Prediction
GraphTCN: Spatio-Temporal Interaction Modeling for Human Trajectory Prediction原创 2020-05-08 13:04:35 · 987 阅读 · 7 评论 -
tensorflow中的tf.pad使用
tensorflow中的tf.pad使用原创 2020-05-07 23:24:32 · 172 阅读 · 0 评论 -
tensorflow 中的tf.clip_by_value()
tensorflow 中的tf.clip_by_value()原创 2020-05-07 18:35:58 · 460 阅读 · 0 评论 -
A Multi-Modal States based Vehicle Descriptor and Dilated Convolutional Social Pooling for Vehicle
A Multi-Modal States based Vehicle Descriptor and Dilated Convolutional Social Pooling for Vehicle原创 2020-05-05 17:22:29 · 227 阅读 · 0 评论 -
A Multi-Modal States based Vehicle Descriptor and Dilated Convolutional Social Pooling for Vehicle
A Multi-Modal States based Vehicle Descriptor and Dilated Convolutional Social Pooling for Vehicle T原创 2020-05-05 16:15:30 · 255 阅读 · 2 评论 -
策略梯度
Policy Gradient Methods for Reinforcement Learning with Function Approximation(PG)在强化学习的算法中存在两种算法,一个是基于价值函数的算法,另一个是基于策略梯度的算法。为什么要提出策略梯度算法呢?基于策略的学习可能会具有更好的收敛性,这是因为基于策略的学习虽然每次只改善一点点,但总是朝着好的方向在改善;...原创 2019-03-08 21:47:56 · 3194 阅读 · 0 评论 -
deepmind论文百度云盘
百度网盘链接:链接:https://pan.baidu.com/s/1S6Vbe1yDis7uXe_LcDylpA提取码:25u0原创 2019-01-13 18:23:57 · 1532 阅读 · 1 评论 -
ImageNet下载网址
http://www.image-net.org/challenges/LSVRC/2012/nonpub-downloads原创 2019-06-20 09:31:09 · 1924 阅读 · 0 评论 -
Python中namedtuple()的使用
Python中namedtuple()的使用,说它是元组,我倒是感觉更像是一个有名字的字典原创 2020-04-26 12:31:17 · 619 阅读 · 0 评论 -
Python中defaultdict()的使用
Python中的defaultdict的使用原创 2020-04-26 10:17:07 · 2212 阅读 · 0 评论 -
graph_nets(deepmind简单的一些注释)
(1)blocks.py# Copyright 2018 The GraphNets Authors. All Rights Reserved.## Licensed under the Apache License, Version 2.0 (the "License");# you may not use this file except in compliance with th...原创 2020-04-26 07:01:12 · 724 阅读 · 0 评论 -
gnn_caphcha(使用deepmind的graphs_net)
(1)curtain.py,读取图片并处理from graph_nets import utils_npfrom pathlib import Pathimport networkx as nximport numpy as npimport randomimport cv2class Code: def __init__(self,path_kind,batch_si...原创 2020-04-25 20:18:02 · 249 阅读 · 0 评论 -
SCALE-Net: Scalable Vehicle Trajectory Prediction Network under Random Number of Interacting Vehicle
简明概要:仅仅描述了主题模型import tensorflow as tffrom tensorflow import kerasfrom tensorflow.keras import layers,initializers,activationsclass EGCN(keras.Model): def __init__(self,f_gcn,f_atten,channe...原创 2020-04-20 12:05:25 · 537 阅读 · 2 评论 -
SCALE-Net: Scalable Vehicle Trajectory Prediction Network under Random Number of Interacting Vehicle
简明概要:仅仅描述了主题模型import torchimport torch.nn as nnimport torch.nn.functional as Fclass EGCN(nn.Module): def __init__(self,f_in,f_gcn,f_atten,channels=4): super(EGCN,self).__init__() self.f_...原创 2020-04-20 10:45:10 · 561 阅读 · 2 评论 -
Tensorflow2(Temporal-level Attention Calculation)
论文 : Vehicle Trajectory Prediction Using LSTMs with Spatial-Temporal Attention Mechanismsimport tensorflow as tffrom tensorflow import kerasfrom tensorflow.keras import initializersclass Te...原创 2020-04-09 19:26:18 · 289 阅读 · 0 评论 -
Pytorch(Temporal-level Attention Calculation)
论文 : Vehicle Trajectory Prediction Using LSTMs with Spatial-Temporal Attention Mechanismsimport torchimport torch.nn as nnimport torch.nn.functional as Fclass TemporalAttention(nn.Module): d...原创 2020-04-09 19:25:41 · 477 阅读 · 0 评论 -
Rezero Linear Pytorch
import torchimport torch.nn as nnclass Rezero_Linear(nn.Module): def __init__(self,inputs,outputs): super(Rezero_Linear,self).__init__() self.inputs = inputs self.outputs = outputs self...原创 2020-04-02 20:59:52 · 232 阅读 · 0 评论 -
Tensorflow2 Repo for "ReZero is All You Need: Fast Convergence at Large Depth"
MultiHeadAttention.pyimport tensorflow as tffrom tensorflow import kerasfrom tensorflow.keras import layers"""class MultiHeadAttention(keras.Model): # https://machinetalk.org/2019/04/29/crea...原创 2020-04-02 11:40:41 · 446 阅读 · 0 评论 -
Graph Attention Pytorch
import torchimport torch.nn as nnclass BatchMultiHeadGraphAttention(nn.Module): # 多头图注意力模型 def __init__(self, n_head, f_in, f_out, attn_dropout, bias=True): super(BatchMultiHeadGraphAttention,...原创 2020-03-12 21:47:36 · 708 阅读 · 0 评论 -
Graph Attention Tensorflow2
import tensorflow as tffrom tensorflow import kerasfrom tensorflow.keras import layers,initializersclass BatchMultiHeadGraphAttention(keras.Model): # 多头图注意力模型 def __init__(self, n_head, f_in, f...原创 2020-03-12 21:43:46 · 192 阅读 · 0 评论 -
tensorflow2 自定义参数变化 并不仅仅是学习率
from tensorflow import kerasstarter_learning_rate = 0.1end_learning_rate = 0.01decay_steps = 10000learning_rate_fn = keras.optimizers.schedules.PolynomialDecay( starter_learning_rate, d...原创 2020-02-06 10:49:34 · 687 阅读 · 0 评论 -
tensorflow2 GUMBEL-SOFTMAX
import tensorflow as tfimport tensorflow_probability as tfp# 生成 OneHotCategoricaldef dist_from_h(h, N, K, z_logit_clip, mode): logits_separated = tf.reshape(h, [-1, N, K]) logits_separated_me...原创 2020-02-06 10:44:43 · 823 阅读 · 0 评论 -
tensorflow2 使用BN Dropout 时的开关
import tensorflow as tffrom tensorflow.keras import layers# 训练阶段with tf.GradientTape() as tape: out = network(x, training=True) # 类似 pytorch 的 model.train()# 在测试阶段,需要设置 training=False, 避免 ...原创 2020-02-04 16:53:28 · 844 阅读 · 2 评论 -
Tensorflow2 使用经典的模型
import tensorflow as tffrom tensorflow.keras import applications"""keras.applications 中共有以下模型DenseNet121(...)DenseNet169(...)DenseNet201(...)InceptionResNetV2(...)InceptionV3(...)MobileNet...原创 2020-02-04 12:11:39 · 567 阅读 · 0 评论 -
sonnet_graph_nets
train.pyimport sonnet as sntimport tensorflow as tffrom dataloader import *from models import *import oscheckpoint_root = "./checkpoints"checkpoint_name = "model"save_prefix = os.path.jo...原创 2019-11-30 23:05:53 · 337 阅读 · 0 评论 -
Python T-sne降维
import csvimport numpy as npfrom sklearn.preprocessing import StandardScalerfrom sklearn import manifoldimport matplotlib.pyplot as pltimport randomimport mathtemp = []with open('qsar_f...原创 2019-11-18 18:08:45 · 377 阅读 · 0 评论 -
python 返回图像中两个矩形区域的交集区域在第一个矩形区域的坐标
def Returns_Coordinates(left_1,right_1,left_2,right_2): # left_1 right_1 block coord = [] left_h = 0 left_w = 0 right_h = 0 right_w = 0 height = min(right_1[0],right_2[0])-max(left_1[0],left...原创 2019-10-21 09:26:21 · 1254 阅读 · 0 评论 -
Python 图像重叠分块
import cv2import math import numpy as npfrom pathlib import Pathfile_name = "121.tif"save_path = './sunxu' # create dir sunxuPath(save_path).mkdir(parents=True, exist_ok=True) # block size...原创 2019-10-19 16:13:49 · 2320 阅读 · 14 评论 -
CAUSAL INDUCTION FROM VISUAL OBSERVATIONS FOR GOAL DIRECTED TASKS
Model_Partsimport torchvisionfrom torch import nnimport torchimport torch.nn.functional as F#create CONV+RELU+POOLclass conv_br(nn.Module): def __init__(self, in_ch, out_ch): super(conv...原创 2019-10-17 22:03:47 · 453 阅读 · 0 评论 -
将CIFAR10数据集拆分为train、val以及test
import cv2import numpy as npfrom pathlib import Pathfrom tqdm import tqdmimport pickle#先读标签with open('cifar-10-batches-py/batches.meta', 'rb') as fo: dict_meta = pickle.load(fo, encoding='b...原创 2019-10-13 11:30:34 · 2578 阅读 · 0 评论 -
将SVHN数据集拆分为train、val以及test
import cv2from scipy.io import loadmatimport numpy as npfrom pathlib import Pathfrom tqdm import tqdm#SVHNdataFile = './oral_image/train_32x32.mat' data = loadmat(dataFile)train_number = ...原创 2019-10-13 11:28:37 · 573 阅读 · 0 评论 -
Python之禅
Python之禅原创 2017-05-08 18:45:56 · 301 阅读 · 0 评论