Closed kobic8 closed 1 year ago
Do you mean using dinov2 or not? As I know it's a self-supervised model and may not that suitable for this task
IDEA-Research did an awesome work building Grounded-DINO which is based on DINO (version1), I was asking if you also plan to build, enrich the grounded-model to DINO-2
IDEA-Research did an awesome work building Grounded-DINO which is based on DINO (version1), I was asking if you also plan to build, enrich the grounded-model to DINO-2
Actually they're different works lol, GroundingDINO is built on DINO from IDEA-Research which has the same name as dino but a detection model
WOW!!! a bit confusing, since this model has the same name as Meta/DINO so for now, the base model of G-DINO has not future work of improvment? are you planning of releasing a code of fine-tuning the model? I have an issue that I want to solve: I have a maritime dataset of boats, jet-skis and other seaborne vehicles. G-DINO detects and labels the targets as boat/ship, but I am interested in refinement to teach the model about the boats' names
WOW!!! a bit confusing, since this model has the same name as Meta/DINO so for now, the base model of G-DINO has not future work of improvment? are you planning of releasing a code of fine-tuning the model? I have an issue that I want to solve: I have a maritime dataset of boats, jet-skis and other seaborne vehicles. G-DINO detects and labels the targets as boat/ship, but I am interested in refinement to teach the model about the boats' names
Very sorry, we did not plan to release the training and fine-tuning code at this time because our paper is still under review and the code is massive and needs to be cleaned, it's a hard job, but we did not have much time to do it now.
I'm going to close this issue now, feel free to reopen it if needed
Hello, are you planning on integrating DINO-2 for faster and more accurate results for bbox+labels?