Just curious: Has anyone tried to directly stream MPEG-G compressed data into your deep learning models— no decompression needed?
Instead of bloated I/O and wasted disk time, one could build smarter:
🧬 Read compressed data ➡️ Transform to tensors ➡️ Train 💡 Lower memory, faster pipeline, greener compute, and bonus would be: It’s what MPEG-G was built for
📦 Let’s redefine what “raw” data means in ML. 💻 If you’re stuck on how to start, ping us—we’ll help!
#MPEGG #MicrobiomeChallenge #EfficientDL
We tried but we found it difficult to even read the mpeg-g file with Genie. Only conversion was made easy. Any tips?