How to create your own dataset and use it to tune the model

longzw1997 / Open-GroundingDino

This is the third party implementation of the paper Grounding DINO: Marrying DINO with Grounded Pre-Training for Open-Set Object Detection.

MIT License

396 stars 62 forks source link

How to create your own dataset and use it to tune the model #22

Closed zy199676 closed 11 months ago

zy199676 commented 11 months ago

May I ask how I can create a dataset for my own data to tune? I am currently working on tasks related to driving area detection and would like to use the weights in the Grounded-SAM model after tuning. Thank you.

BIGBALLON commented 11 months ago

For training, we use the odvg data format to support both OD data and VG data. Before model training begins, you need to convert your dataset into odvg format, see data_format.md | datasets_mixed_odvg.json | coco2odvg.py | grit2odvg for more details.

For testing, we use coco format, which currently only supports OD datasets.

zfr00 commented 8 months ago

sorry to ask can you supports the VG datasets now?

BIGBALLON commented 8 months ago

sorry to ask can you supports the VG datasets now?

Sorry, we currently lack the bandwidth to support additional reviews and do not have plans to extend our support. If you find the time, you are welcome to submit a PR to support this feature. 😄