Faster image pipelines · FluxML/FastAI.jl

(0 comentário) (0 reação) (0 responsável)Julia (49 forks)batch import

enhancementhelp wanted

Métricas do repositório

Some low-hanging fruit that FastAI.jl can pick to become faster here:

builtin presizing support
- prototype at https://github.com/lorenzoh/DataBlockStorage.jl/
- using JpegTurbo.jl's downsized loading so no storing to disk is needed (https://github.com/JuliaIO/JpegTurbo.jl/issues/15#issuecomment-1031548344)
use inplace operations
- DataAugmentation.jl already supports inplace operations, FastAI.jl just needs to add a buffered encode! interface and use those paths
- JpegTurbo.jl image loading could also be done inplace (https://github.com/JuliaIO/JpegTurbo.jl/issues/23)
- the whole pipeline could become non-allocating
use MLUtils.jl (#196)
- new DataLoader seems more stable and possibly a little faster
run projective transformations on GPU (https://github.com/lorenzoh/DataAugmentation.jl/issues/48)

Direção de pesquisa: Concentre se em implementar o suporte para redimensionamento prévio usando o carregamento reduzido do JpegTurbo.jl, seguindo o protótipo no DataBlockStorage.jl. Teste a integração e faça benchmark das melhorias de velocidade.
Pilha de tecnologia: julia
Domain: machine learning
Tipo Issue: Performance
Difficulty: 2
Tempo estimado: Meio dia
Status da atividade: Recente
Clarity: Claro
Prerequisites: JuliaGitFastAI.jl
Simpatia para novatos: 70