You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
I'm drawn to the OpenMMLab framework suite, especially for model development in the CV and VLP domains.
I'm currently developing a model designed for a detection task combined with a high-granularity multimodal classification task. As such, both mmdet and mmpretrain appear relevant due to their respective toolsets.
I'm seeking guidance on best practices for developing parts of my model using mmdet while simultaneously developing other parts of my model with mmpretrain.
My present challenge is with the setup.cfg which uses symbolic links for the recommended installation (pip install -e .) from the source. When I'm in the mmdet directory and want to import an editable module from mmpretrain, which is concurrently under development, I encounter an issue. From the mmdet directory, mmpretrain seems nonexistent due to the symbolic link installation. It's only accessible within its own directory.
Is there a streamlined approach to develop with both modules in editable mode? While I've consulted the documentation, I couldn't find any specific guidelines on this topic. I came across the mmpose + open_detection example in the playground repository, which integrates mmpose with mmdet. However, it doesn't seem to align with my use case of jointly developing with mmdet and mmpose.
I've considered appending the Python path variable with the directories of both mmdet and mmpretrain, although I haven't tested this yet. However, directly modifying the Python path might not be the most recommended approach. I'd appreciate insights or alternative suggestions for my dual development scenario.
reacted with thumbs up emoji reacted with thumbs down emoji reacted with laugh emoji reacted with hooray emoji reacted with confused emoji reacted with heart emoji reacted with rocket emoji reacted with eyes emoji
-
Hello,
I'm drawn to the OpenMMLab framework suite, especially for model development in the CV and VLP domains.
I'm currently developing a model designed for a detection task combined with a high-granularity multimodal classification task. As such, both mmdet and mmpretrain appear relevant due to their respective toolsets.
I'm seeking guidance on best practices for developing parts of my model using mmdet while simultaneously developing other parts of my model with mmpretrain.
My present challenge is with the setup.cfg which uses symbolic links for the recommended installation (pip install -e .) from the source. When I'm in the mmdet directory and want to import an editable module from mmpretrain, which is concurrently under development, I encounter an issue. From the mmdet directory, mmpretrain seems nonexistent due to the symbolic link installation. It's only accessible within its own directory.
Is there a streamlined approach to develop with both modules in editable mode? While I've consulted the documentation, I couldn't find any specific guidelines on this topic. I came across the mmpose + open_detection example in the playground repository, which integrates mmpose with mmdet. However, it doesn't seem to align with my use case of jointly developing with mmdet and mmpose.
I've considered appending the Python path variable with the directories of both mmdet and mmpretrain, although I haven't tested this yet. However, directly modifying the Python path might not be the most recommended approach. I'd appreciate insights or alternative suggestions for my dual development scenario.
Thank you!
Beta Was this translation helpful? Give feedback.
All reactions