Jianyuan Wang

Results 238 comments of Jianyuan Wang

What do you mean by local features? Given the size you mentioned there, do you want a CxH_localxW_local feat map? For example, if you want features from NetVLAD, you can...

``` from hloc.utils.database import COLMAPDatabase db = COLMAPDatabase.connect(database) ``` More details: https://github.com/cvg/Hierarchical-Localization/blob/e3e953f4db00c3b9b14951482349d5ddd9424452/hloc/triangulation.py#L36-L61

You can check COLMAP https://github.com/colmap/colmap/blob/a36156a64191127f83f6011616044d889667209c/src/colmap/sensor/models.h#L224 for its definition. k is the radial distortion parameter ``` // Simple camera model with one focal length and one radial distortion // parameter. //...

Hi @hsmmoj , how many images do you have?

Colmap expects you to run sparse reconstruction first to get sparse 3d point clouds and camera poses, and then can run dense reconstruction. If you don't want to run sparse...

Hi @antithing , feel free to check https://github.com/facebookresearch/vggsfm

Hi. I tried to reproduce this but it seems working correctly for me. I can dig into this further though chunking may give you something good at the stage. Another...

Hi @booker-max , Your proposed solutions look great. I would prefer the second option as it does not need to finetune a model. For the first one, you may need...

Hi I guess it is due to the cropping methods, for example, something related to this: https://github.com/facebookresearch/vggt/pull/57 Would you mind sharing the whole image set so i can have a...