Zero_shot_action_reco_STGCN

KO | EN

gitlite — search

#llm #golang #android #linux #nodejs #macos #java #awesome #cli #kubernetes #vue #pytorch #security #openai

Zero_shot_action_reco_STGCN

(★ 16)

Zero shot action recognition using spatial temporal graph convolutional networks

Zero_shot_action_reco_STGCN 최신버젼 다운로드

최종 버전 다운로드 (.zip)

// repository documentation

Was this content helpful?

★ 0 (0 ratings)

# Zero_shot_action_reco_STGCN

Zero shot action recognition using Spatial Temporal Graph Convolutional Network and Deep Visual-Semantic Embedding Model

To generate embeddings for action class names:

- Download fasttext: (clone the repo)
      https://github.com/facebookresearch/fastText

- Download pretrained skip-gram model
      https://drive.google.com/file/d/0B6VhzidiLvjSaER5YkJUdWdPWU0/view

More pretrained models available here: 
      https://github.com/epfml/sent2vec

- Navigate to the fasttext folder and run the fasttext binary.
      Provide your downloaded model as a .bin and your sentences as a txt file

$ ./fasttext print-sentence-vectors model.bin < text.txt > outfile.txt
      $ ./fasttext print-sentence-vectors wiki_bigrams.bin < action_classes_original.txt > output_embeddings.txt

This will save embeddings to the output txt file
  
- Run generate_embeddings() in neighbours.py to get these embeddings as a .npy

UNSEEN CLASSES

Hand picked   = [3,8,11,16,51]

NEAREST NEIGH = [14,16,43,49,58]

FURTHER NEIGH = [10,24,41,52,55]

Note: The classes go from  0 to 59

Zero_shot_action_reco_STGCN

Zero_shot_action_reco_STGCN 최신버젼 다운로드

오늘의 추천 프로젝트

Was this content helpful?

Recent Feedback

Zero_shot_action_reco_STGCN

Zero_shot_action_reco_STGCN 최신버젼 다운로드

오늘의 추천 프로젝트

Was this content helpful?

Recent Feedback

Download README