- 博客(155)
- 收藏
- 关注
RSS订阅
转载 ***CV AND CNN 总结***
转自： https://zhuanlan.zhihu.com/p/31116104，侵删！本人入门CV方向时追着看无数的公开课，教程，入门指南之类的，每每发现别人共享的资料，百十来本书，百十个视频公开课连接，力不从心底学了几天，发现都是白扯。看N遍CS231n，读paper，精读这个领域的100篇paper，写review，写代码，写自己的想法，写自己与别人的差距。于是乎重新整理知乎专栏，把...
2018-10-27 11:03:37
449
原创 算法与数据学习知识点
常见的时间复杂度：O(1) < O(logn) < O(n) < O(nlogn) < O(n^2) < O(n^3) < O(2^n) < O(n!) < O(n^n)
2020-10-29 09:00:44
215
转载 GAN生成数据-数据扩增
http://www.zhuanzhi.ai/document/7fd3ab2d82d75165eff768a06dfb74bf ***一文概览用于数据集增强的对抗生成网络架构https://github.com/AntreasAntoniou/DAGAN ***DAGAN: Data Augmentation Generative Adversarial Networkshttps:...
2019-07-11 09:28:53
9524
转载 目标检测未来方向
https://www.zhihu.com/question/280703314/answer/564235579 ***目标检测领域还有什么可以做的？zero shot, one shot,Incremental Learning
2019-06-21 11:44:59
2638
转载 pytroch
https://www.jiqizhixin.com/articles/2019-06-02-4 ***万字综述，核心开发者全面解读PyTorch内部机制https://www.jiqizhixin.com/articles/2019-06-18-10 ***PyTorch代码调试利器: 自动print每行代码的Tensor信息...
2019-06-18 15:45:03
251
转载 数据不平衡
https://www.jiqizhixin.com/articles/2019-06-18-5 ***极端类别不平衡数据下的分类问题研究综述
2019-06-18 15:38:14
200
转载 learning rate schedule
https://medium.com/@sachinmehta.ngb/faster-training-of-efficient-cnns-657953aa080 ***Faster Training for Efficient CNNscosine learning rate
2019-06-09 16:37:21
1120
转载 强化学习库
https://medium.com/data-from-the-trenches/choosing-a-deep-reinforcement-learning-library-890fb0307092 ***库比较KerasRL/TensorForce/OpenAI Baselines/Stable Baselines/TF Agents
2019-05-31 09:31:32
592
转载 乐透理论
https://www.technologyreview.com/s/613514/a-new-way-to-build-tiny-neural-networks-could-create-powerful-ai-on-your-phone/ ***网络训练之前找到最好配置的小模型节省算力训练，阐述人们为什么常常选择大模型，来源于THE LOTTERY TICKET HYPOTHESIS ...
2019-05-29 20:09:39
370
转载 small目标检测
https://zhuanlan.zhihu.com/p/60033229 ***提升小目标检测的数据增量策略，上采样增加包含小物体的图片样本，小物体样本复制扩增；定义小物体尺寸（小于32*32）；分析coco数据集小物体样本不均衡；https://www.jiqizhixin.com/articles/2018-05-21-9 ***增加检测类别？这是一份目标检测的基础指南...
2019-05-29 17:29:03
213
转载 33个神经网络技巧
转自机器之心，侵删！33个神经网络技巧特斯拉人工智能部门主管 Andrej Karpathy 发布新博客，介绍神经网络训练的技巧。Andrej Karpathy 是深度学习计算机视觉领域、生成式模型与强化学习领域的研究员。博士期间师从李飞飞。在读博期间，两次在谷歌实习，研究在 Youtube 视频上的大规模特征学习，2015 年在DeepMind实习，研究深度强化学...
2019-05-25 10:39:49
475
转载 对象检测小型化-thunderNet
https://blog.youkuaiyun.com/u011344545/article/details/88927579 https://blog.youkuaiyun.com/shanglianlm/article/details/88932983
2019-05-24 14:45:16
195
转载 tf objective det api
https://www.cnblogs.com/HaijunLv/p/9101957.html ***细粒度分析https://github.com/EdjeElectronics/TensorFlow-Object-Detection-API-Tutorial-Train-Multiple-Objects-Windows-10#7-export-inference-graph ***...
2019-05-10 15:49:51
122
转载 tf load ssd pb
https://blog.youkuaiyun.com/sxlsxl119/article/details/81937815 ***TensorFlow保存训练模型为pb文件，读取pb恢复模型，C++调用pb模型https://blog.youkuaiyun.com/u014432647/article/details/75276718 ***如何用Tensorflow训练模型成pb文件和和如何加载已经训练好...
2019-04-23 11:32:57
239
转载 文本分类
https://medium.com/jatana/report-on-text-classification-using-cnn-rnn-han-f0e887214d5f ***Report on Text Classification using CNN, RNN &HAN
2019-04-22 14:30:51
140
转载 proto 存参数/load 参数
通过protobuff初始化网络参数，读取prototxt：https://blog.youkuaiyun.com/flyan338/article/details/8448518https://blog.youkuaiyun.com/pc153262603/article/details/79854323
2019-04-10 17:19:51
373
转载 Post
2019.3.30***This post studies a similar problem: suppose each color channel has a different size.The best thing to do in this case is to just downscale the pictures to match the smallest channel di...
2019-03-30 10:01:41
212
转载 protobuff
https://github.com/halfrost/Halfrost-Field/blob/master/contents/Protocol/Protocol-buffers-encode.md#%E5%85%AD-protocol-buffer-%E7%BC%96%E7%A0%81%E5%8E%9F%E7%90%86 ***高效的数据压缩编码方式 Protobufhttps://ww...
2019-03-08 10:32:20
514
转载 tensorflow c++
http://mathmach.com/2017/10/09/tensorflow_c++_api_prediction_first/ ***使用TensorFlow C++ API构建线上预测服务 - 篇1http://mathmach.com/2017/10/22/tensorflow_c++_api_prediction_third/ ***使用TensorFlow C++ AP...
2019-03-07 18:57:42
296
1
转载 ckpt转为pb
https://blog.youkuaiyun.com/guyuealian/article/details/82218092 ***tensorflow实现将ckpt转pb文件
2019-02-15 11:26:40
1277
转载 ROS and robot control
ROShttps://pojenlai.wordpress.com/ ROS系列文整理 (ROS Tutorials)https://pojenlai.wordpress.com/2017/12/02/%E4%B8%80%E8%B5%B7%E8%AE%80-end-to-end-training-of-deep-visuomotor-policies/ 深入淺出 End-to-En...
2019-02-01 15:19:18
185
转载 k8s落地
https://www.jiqizhixin.com/articles/2019-01-31-14 Yarn已过时！Kubeflow实现机器学习调度平台才是未来
2019-01-31 16:00:49
390
转载 合并PB文件
https://blog.youkuaiyun.com/qq26983255/article/details/85797707 https://blog.youkuaiyun.com/mogoweb/article/details/83064819
2019-01-31 09:18:34
735
转载 有用的blog
转载文章，著作权归作者所有。如有不妥之处，请联系删帖。2018 paper:https://github.com/amusi/daily-paper-computer-vision108关键点人脸识别：https://github.com/zeusees/HyperLandmarkvm配置：http://blog.youkuaiyun.com/u013142781/article/details/5...
2019-01-31 08:55:22
1308
转载 CNN结构
https://www.jiqizhixin.com/articles/2019-01-25-6 ***7大类深度CNN架构创新综述https://image.jiqizhixin.com/uploads/editor/2e2ad4fe-b45f-4f03-ac93-56b1246da60a/640.jpegA Survey of the Recent Architectures of...
2019-01-25 11:42:14
232
转载 数据预处理&pandas
https://towardsdatascience.com/why-and-how-to-use-pandas-with-large-data-9594dda2ea4c ***Why and How to Use Pandas with Large Data https://www.jiqizhixin.com/articles/2019-01-22-5 ***还在为数据清洗抓狂...
2019-01-25 10:58:13
154
转载 轨迹预测
https://medium.com/analytics-vidhya/game-of-modes-diverse-trajectory-forecasting-with-pushforward-distributions-315b1b30d5e6 ***Game of Modes: Diverse Trajectory Forecasting with Pushforward Distrib...
2019-01-15 11:30:37
7121
转载 RL2_policy_gradients_mainly
https://flyyufelix.github.io/2017/10/12/dqn-vs-pg.html ***Deep Q Network vs Policy Gradients - An Experiment on VizDoom with Kerashttp://karpathy.github.io/2016/05/31/rl/ ***Deep Reinforcement...
2019-01-14 14:36:37
229
转载 迁移学习
https://zhuanlan.zhihu.com/p/38425434 ***2018迁移学习顶峰https://developer.nvidia.com/transfer-learning-toolkit ***英伟达迁移学习tool kits，sensor adaption，model prunehttps://mp.weixin.qq.com/s?__biz=MzI3OD...
2019-01-01 15:46:57
246
转载 autopilot_donkey car
https://flyyufelix.github.io/2018/09/11/donkey-rl-simulation.html ***Train Donkey Car in Unity Simulator with Reinforcement Learning 牛人博客 http://www.donkeycar.com/ ***Donkey Carhttps://githu...
2018-12-27 17:34:30
608
转载 RL_1_readings_dqn_mainly
牛逼的RL源https://simoninithomas.github.io/Deep_reinforcement_learning_Course/#syllabus https://medium.com/@jonathan_hui/rl-dqn-deep-q-network-e207751f7ae4 https://spinningup.openai.com/en/latest/u...
2018-12-25 16:00:06
387
转载 OCR
https://zhuanlan.zhihu.com/p/37678860 ***【领域报告】图像OCR年度进展|VALSE2018之十一
2018-12-18 16:49:00
161
转载 door_inside
https://www.cnblogs.com/rubbninja/tag/%E5%AE%A4%E5%86%85%E5%AE%9A%E4%BD%8D/ https://machinelearningmastery.com/indoor-movement-time-series-classification-with-machine-learning-algorithms/ ***Ind...
2018-12-17 11:05:51
132
转载 BN caffe 源码实现
https://xmfbit.github.io/2018/01/08/caffe-batch-norm/另，通过阅读源码，nvcaffe中batch_norm_param中如果没有设置scale_filler或者bias_filler，则gamme默认初始化为1.0，bate默认初始化为0.0https://blog.youkuaiyun.com/j2IaYU7Y/article/det...
2018-12-10 09:55:23
496
转载 keras 数据读取
https://github.com/JustinhoCHN/fancy_keras/blob/master/script/data_input_fancy.py ***FancyKeras-数据的输入（花式）https://zhuanlan.zhihu.com/p/32679425 ***FancyKeras-数据的输入（花式） ...
2018-11-30 16:54:58
667
转载 机器视觉数据增强
https://mp.weixin.qq.com/s?__biz=MzI1NTE4NTUwOQ==&mid=2650325586&idx=1&sn=69a7e8482d884dda869290581a50a6d6&chksm=f235a558c5422c4eabe75f6db92fd13a3ca662d648676461914c9bfd1f4e22affe12430...
2018-11-28 09:07:39
314
转载 mpi
https://blog.youkuaiyun.com/yanxiangtianji/article/details/55255318 ***MPI选择错误的网卡/网段的解决方案https://docs.nvidia.com/deeplearning/sdk/nccl-developer-guide/index.html#ncclknobs https://www.open-mpi.org/...
2018-10-31 14:46:52
298
转载 人体属性
https://craigie1996.github.io/2018/05/11/Pedestrian-Attribute-Recognition-%E8%B0%83%E7%A0%94%E7%AC%94%E8%AE%B0/ ***Pedestrian Attribute Recognition 调研笔记，以保存为PDF
2018-10-24 10:01:54
735
1
空空如也
空空如也
TA创建的收藏夹 TA关注的收藏夹
TA关注的人