-
Hi, thank you very much for your great work!
I want to transfer the COCO pretrain to my own satellite images, which are high resolution bird views.
I first try the pretrain weight provided by MagicL…
-
Here is the plan for December, and we will release v0.10.0 in the end of the month. Our focus in this iteration is to enrich the supported tasks. We will add support for 2d face keypoint localization.…
-
[2022-05-31 02:44:25] [316241c6] [30] urllib3.connectionpool: http://192.168.2.109:8080 "GET /api/v1/allocations/7.ee72787d-c20e-4824-b0da-ef3e3287e708.1/signals/preemption?timeout_seconds=60 HTTP/1…
-
Hi @mihaidusmanu, thanks so much for releasing the codes and implementation as well as evaluation metrics, and supports to make this repository very complete. I find the performance and the evaluation…
-
Hi.
Thanks for opening the code.
I trained the model as README and just increased batch size from 8(default) to 32.
I used 2 GPUs using dataparallel to use multi-GPU and it took 2 days to comple…
-
-
Hi, thanks for the great project for building VL pipeline. It is really helpful for my research.
I want to ask some questions for reproducing results of Aachen day night v1.1.
I'm using off-the-s…
-
## ❓ Questions and Help
When training the model on keypoints I noticed that you need to add keypoint_names and keypoint_flip_map as key value pairs respectively in the dictionary in MetadataCatalog…
-
Hi,
I followed your direction to run 3D reconstruction on south-building dataset. I got the mean_reproj_error: 1.400.
However, when I rebulit the same scene using COLMAP with default settings, I …
-
Hi, @jatentaki
Thank you for your great work! I have some further questions:
1. have you tested disk's result on aachen localization benchmark and ETH benchmark (for 3D construction tasks)? How's…