– 01 –
Face Recognition
# The world’s simplest face recognition library
This project claims to be the world’s simplest face recognition library, which can be called using Python and the command line. Constructed using dLIB’s state-of-the-art deep learning face recognition technology, the library achieved 99.38% accuracy on Labeled Faces in the Wild Benchmark.
Project links:
https://github.com/ageitgey/face_recognition
– 02 –
MUSE
# Multilingual word vector Python library
The Python library for Multilingual word vectors, open source by Facebook, provides fastText-based multilingual word vectors and large-scale high-quality bilingual dictionaries, both unsupervised and supervised. Where supervised methods use a bilingual dictionary or the same string, unsupervised methods do not use any parallel data.
For details of unsupervised methods, please refer to the paper Word Translation without Parallel Data.
Paper Links:
https://www.paperweekly.site/papers/1097
Project links:
https://github.com/facebookresearch/MUSE
– 03 –
FoolNLTK
# Chinese processing kit
Features of the project:
• May not be the fastest open source Chinese word segmentation, but it is probably the most accurate open source Chinese word segmentation
• Based on BiLSTM model training
• Include word segmentation, part-of-speech tagging, entity recognition, all have relatively high accuracy
• User-defined dictionaries
Project links:
https://github.com/rockyzhengwu/FoolNLTK
– 04 –
Arnold
# The best GAME AI at Playing Doom
This project, from Carnegie Mellon University, is the PyTorch open-source code for Arnold, the 2017 Doom AI deathmatch champion of VizDoom.
Paper Links:
https://www.paperweekly.site/papers/1440
Project links:
https://github.com/glample/Arnold
– 05 –
Bottom-Up Attention VQA
#2017 VQA Challenge first prize
This project is a PyTorch reproduction of two papers from the first place team of the 2017 VQA Challenge.
Thesis: Bottom-up and top-down Attention for Image Captioning and Visual Question appeal
Link: https://www.paperweekly.site/papers/754
Thesis: Tips and Tricks for Visual Question Answering: Learnings from the 2017 Challenge
Link: https://www.paperweekly.site/papers/1441
Project links:
https://github.com/hengyuan-hu/bottom-up-attention-vqa
– 06 –
YOLOv2 – PyTorch
# YOLOv2 PyTorch version
A PyTorch version of the famous object detection library YOLOv2, this project can also convert trained Models into adapters for Caffe 2.
Project links:
https://github.com/ruiminshen/yolo2-pytorch
– 07 –
Simple Railway Captcha Solver
# Taiwan Railway booking verification code identification based on CNN
The Convolutional Neural Network is used to implement the verification codes of the Taiwan Railway ticket booking website. Parts of the training set are generated by imitating the pattern of the verification codes, and parts of the verification set are extracted from the Taiwan Railway ticket booking website, and about 1000 marks are manually made.
At present, the single-code identification rate of the verification set for the 6-code type verification code reaches 98.84%, and the overall identification success rate reaches 91.13%.
Project links:
https://github.com/JasonLiTW/simple-railway-captcha-solver
– 08 –
AlphaZero-Gomoku
# Play gobang with AlphaZero
This is an implementation of the AlphaZero algorithm applied to backgammon, which is much simpler than go or chess, so it takes just a few hours to train a decent AI model.
AlphaZero: Mastering Chess and Shogi by Self-play with a General Reinforcement Learning Algorithm
Link: https://www.paperweekly.site/papers/1297
AlphaGo Zero: Mastering the Game of Go without human knowledge
Link: https://www.paperweekly.site/papers/942
Project links:
https://github.com/junxiaosong/AlphaZero_Gomoku
– 09 –
gym-extensions
#OpenAI Gym Extension set
This is an extension package of OpenAI Gym library, which realizes multi-task learning, transfer learning, inverse enhancement learning and other functions.
Project links:
https://github.com/Breakend/gym-extensions
– 10 –
Myia
#Python Deep learning framework
Myia is a new Python deep learning framework featuring ease of use, automatic differentiation, and performance optimization.
The original post was published on January 02, 2018
This article is from the cloud community partner “Datapai THU”. For relevant information, you can follow the wechat public account of “Datapai THU”