Excellent work! It would be great to compare the performance of different local features and matchers.

Can I add the SFD2 - semantic local feature (https://github.com/feixue94/sfd2) and IMP - iterative matching matcher (https://github.com/feixue94/imp-release) to the repository?

🐛 Bug Report

1. As in the alike.py in hloc
   From alike import ALike as Alike_, from alike import configs are not recognized by alike
   2.And in the d2net algorithm
   From lib. model_test import D2Net as _D2Net
   From lib.pyramid import process_multiscale
   model_test and pyramid

New to artificial intelligence, request a reply

I'm using Windows and WSL, but I keep getting this error and can't really get it to work.

How do I fix this error please?

Task exception was never retrieved
future: <Task finished name='j7l5nczw6pi_3' coro=<Queue.process_events() done, defined at /home/qf/miniconda3/envs/imw/lib/python3.10/site-packages/gradio/queueing.py:342> exception=1 validation error for PredictBody
event_id
  Field required [type=missing, input_value={'data': ['data:image/jpe...on_hash': 'j7l5nczw6pi'}, input_type=dict]
    For further information visit https://errors.pydantic.dev/2.5/v/missing>
Traceback (most recent call last):
  File "/home/qf/miniconda3/envs/imw/lib/python3.10/site-packages/gradio/queueing.py", line 346, in process_events
    client_awake = await self.gather_event_data(event)
  File "/home/qf/miniconda3/envs/imw/lib/python3.10/site-packages/gradio/queueing.py", line 219, in gather_event_data
    data, client_awake = await self.get_message(event, timeout=receive_timeout)
  File "/home/qf/miniconda3/envs/imw/lib/python3.10/site-packages/gradio/queueing.py", line 448, in get_message
    return PredictBody(**data), True
  File "/home/qf/miniconda3/envs/imw/lib/python3.10/site-packages/pydantic/main.py", line 164, in __init__
    __pydantic_self__.__pydantic_validator__.validate_python(data, self_instance=__pydantic_self__)
pydantic_core._pydantic_core.ValidationError: 1 validation error for PredictBody
event_id
  Field required [type=missing, input_value={'data': ['data:image/jpe...on_hash': 'j7l5nczw6pi'}, input_type=dict]
    For further information visit https://errors.pydantic.dev/2.5/v/missing

Downloading checkpoints/PointModel_v0.pth (34 MB)
Error downloading object: checkpoints/PointModel_v0.pth (17c1adf): Smudge error: Error downloading checkpoints/PointModel_v0.pth (17c1adfc8c22b044a9538019101bae0740a9d054c1b4fecd80b52d642272b9ff): batch response: This repository is over its data quota. Account responsible for LFS bandwidth should purchase more data packs to restore access.

Errors logged to 'C:\Users\Administrator\Project\ORS2SEM\data\实验4\image-match\image-matching-webui.git\modules\third_party\lanet\lfs\logs\20240315T102550.0545338.log'.
Use git lfs logs last to view the log.
error: external filter 'git-lfs filter-process' failed
fatal: checkpoints/PointModel_v0.pth: smudge filter lfs failed
fatal: Unable to checkout '58ec4881f43fa082fbd18a61477224266bdc111c' in submodule path 'third_party/lanet'

Congrats on the project. Recently I played with and I like it. What I found interesting is that it would be nice for someone to estimate a fundamental matrix or a homography between the two images thus I added the necessary functionalities leveraging OpenCV. If you find this useful please let me know.

I believe that there is an issue there :

It should be

rectified_image1 = cv2.warpPerspective(
                img1, H, (img0.shape[1], img0.shape[0])
            )

Otherwise the wrapped image has the wrong size.

need to source? I met this question when run main.py
File "image-matching-webui-main/hloc/matchers/sold2.py", line 11, in
from sold2.model.line_matcher import LineMatcher

Looks like you're using the hloc codebase here. Is your code a good example of how to integrate RoMa into hloc to improve the quality of hloc results?

hey great repo , just a question can you shed any light on which techniques gave the best results (lower FPs and better accuracies)

Hi, thanks for posting this great code
Please add COTR and Quadtree as well
Thank you very much

bug:
how can I solve this?

(imw) maomao@maomao415:~/Codes/image-matching-webui$ python ./app.py Traceback (most recent call last): File "/home/maomao/Codes/image-matching-webui/./app.py", line 2, in <module> import gradio as gr File "/home/maomao/anaconda3/envs/imw/lib/python3.10/site-packages/gradio/__init__.py", line 3, in <module> import gradio.components as components File "/home/maomao/anaconda3/envs/imw/lib/python3.10/site-packages/gradio/components/__init__.py", line 1, in <module> from gradio.components.annotated_image import AnnotatedImage File "/home/maomao/anaconda3/envs/imw/lib/python3.10/site-packages/gradio/components/annotated_image.py", line 12, in <module> from gradio import utils File "/home/maomao/anaconda3/envs/imw/lib/python3.10/site-packages/gradio/utils.py", line 353, in <module> class AsyncRequest: File "/home/maomao/anaconda3/envs/imw/lib/python3.10/site-packages/gradio/utils.py", line 372, in AsyncRequest client = httpx.AsyncClient() File "/home/maomao/anaconda3/envs/imw/lib/python3.10/site-packages/httpx/_client.py", line 1395, in __init__ proxy_map = self._get_proxy_map(proxies, allow_env_proxies) File "/home/maomao/anaconda3/envs/imw/lib/python3.10/site-packages/httpx/_client.py", line 216, in _get_proxy_map return { File "/home/maomao/anaconda3/envs/imw/lib/python3.10/site-packages/httpx/_client.py", line 217, in <dictcomp> key: None if url is None else Proxy(url=url) File "/home/maomao/anaconda3/envs/imw/lib/python3.10/site-packages/httpx/_config.py", line 336, in __init__ raise ValueError(f"Unknown scheme for proxy URL {url!r}") ValueError: Unknown scheme for proxy URL URL('socks://127.0.0.1:7891/')

This is a great webui, but kinda useful.

Is there/will be in future option to export matching data to colmap (or other similar programs) and batch processing option for using this webui in photogrammetry?

*** Aborted at 1707060306 (unix time) try "date -d @1707060306" if you are using GNU date ***
@ 0x7ffdc24f4090 log2f
@ 0x7ff7812c1f5c OPENSSL_Applink
@ 0x7ffdb903f73f __C_specific_handler
@ 0x7ffdc4b1441f __chkstk
@ 0x7ffdc4a8e466 RtlFindCharInUnicodeString
@ 0x7ffdc4b1340e KiUserExceptionDispatcher
@ 0x7ffd6e1f264e void __cdecl __ExceptionPtrRethrow(void const * __ptr64)
@ 0x7ffc2f2b38de public: void __cdecl c10::ivalue::Future::markCompleted(void) __ptr64
@ 0x7ffc2f531e19 struct _object * __ptr64 __cdecl THPGenerator_initDefaultGenerator(struct at::Generator)
@ 0x7ffc2a70dea5 (unknown)
@ 0x7ffc2ac66b29 PyInit_pycolmap
@ 0x7ffd9ade1080 (unknown)
@ 0x7ffd9ade2715 __NLG_Return2
@ 0x7ffdc4b13c66 RtlCaptureContext2
@ 0x7ffc2a7109b9 (unknown)
@ 0x7ffd5d49fd0d PyCFunction_GetFlags
@ 0x7ffd5d45acf8 _PyObject_MakeTpCall
@ 0x7ffd5d464141 PyComplex_AsCComplex
@ 0x7ffd5d4a43e4 _PyObject_GenericGetAttrWithDict
@ 0x7ffd5d4a3d18 PyObject_GetAttr
@ 0x7ffd5d4a3969 PyObject_GetAttrString
@ 0x7ffc2a74609e PyInit_pycolmap
@ 0x7ffc2a79dc6d PyInit_pycolmap
@ 0x7ffc2a77651d PyInit_pycolmap
@ 0x7ffc2a734170 (unknown)
@ 0x7ffc2a740a48 PyInit_pycolmap
@ 0x7ffc2a74054e PyInit_pycolmap
@ 0x7ffd5d5a0db9 PyImport_AppendInittab
@ 0x7ffd5d5a074c PyImport_Import
@ 0x7ffd5d49f919 PyCFunction_GetFlags
@ 0x7ffd5d56f8fc PyEval_GetFuncDesc
@ 0x7ffd5d56a720 _PyEval_EvalFrameDefault
May I ask if you know the reason? Thank you very much

Hello, and thanks for your great work!

I am having errors whenever trying to use any combination of LightGlue.
I.e.
File "/home/mikkel/image-matching-webui/hloc/matchers/../../third_party/LightGlue/lightglue/lightglue.py", line 349, in forward
kpts0, kpts1_ = data0['keypoints'], data1['keypoints']
IndexError: too many indices for tensor of dimension 4

Hope you can help

Hello,
I get a lot of ResolvePackageNotFound errors from the conda environment.
Could you please regenerate the environment.yaml so to make sure it is up to date?
Thanks!

CONDA_RESTORE_FREE_CHANNEL=1 conda env create --name imw --file environment.yaml
Collecting package metadata (repodata.json): done
Solving environment: failed

ResolvePackageNotFound: 
  - pyrsistent==0.18.1=py310he2412df_1
  - jpeg==9e=h8ffe710_1
  - m2w64-libwinpthread-git==5.0.0.4634.697f757=2
  - sip==6.5.1=py310h8a704f9_2
  - pywin32==303=py310he2412df_0
  - protobuf==3.19.1=py310hd77b12b_0
  - libbrotlidec==1.0.9=h8ffe710_7
  - jupyter==1.0.0=py310haa95532_7
  - m2w64-gcc-libgfortran==5.3.0=6
  - libglib==2.70.2=h3be07f2_4
  - pyqt==5.15.4=py310h8a704f9_0
    ...

after clone this repository , could not find the file .

model_zoo in file extra_utils/utils.py.

Hi,

I was using LoFTR but I was unable to obtain the confidence score. Instead it always returned [1 1 1 1 ...].

I believe there is an issue there, and maybe it is the same for sparse detectors too :
https://github.com/Vincentqyw/image-matching-webui/blob/f87ed7aad49842ab0c542c220356dbeb12d717d1/hloc/match_dense.py#L373C1-L376C67

Indeed, loftr prediction function does not return score under the key "mconf", but under the key "scores", hence I modified the piece of code by taking into account the "scores" key :

        if "mconf" in pred.keys():
            ret["mconf"] = pred["mconf"].cpu().numpy()
        elif "scores" in pred.keys() :
            ret["mconf"] = pred["scores"].cpu().numpy()
        else:
            ret["mconf"] = np.ones_like(kpts0.cpu().numpy()[:,0])

And now I can retrieve the confidence score.

"Scores" works for LoFTR but I dit not try other detectors.

python app.py
*** Aborted at 1711654442 (unix time) try "date -d @1711654442" if you are using GNU date ***
@ 0x7ff8e89a0ef5 _seh_filter_exe
@ 0x7ff75abe1cfc (unknown)
@ 0x7ff8ca31ecd0 __C_specific_handler
@ 0x7ff8eae323af __chkstk
@ 0x7ff8eade14b4 RtlRaiseException
@ 0x7ff8eae30ebe KiUserExceptionDispatcher
@ 0x7ff8bf3f25ee void __cdecl __ExceptionPtrRethrow(void const * __ptr64)
@ 0x7ff85414135f (unknown)
@ 0x7ff85414393c (unknown)
@ 0x7ff84f92dea5 (unknown)
@ 0x7ff84fe86b29 PyInit_pycolmap
@ 0x7ff8c1171080 (unknown)
@ 0x7ff8c11726a5 __NLG_Return2
@ 0x7ff8eae31716 RtlCaptureContext2
@ 0x7ff84f9309b9 (unknown)
@ 0x7ff85660c24a PyObject_IsTrue
@ 0x7ff8565d60db _PyObject_MakeTpCall
@ 0x7ff856740045 _Py_gitversion
@ 0x7ff8565d946c _PyObject_GenericGetAttrWithDict
@ 0x7ff8565d8967 PyObject_GetAttr
@ 0x7ff856618831 PyObject_GetAttrString
@ 0x7ff84f96609e PyInit_pycolmap
@ 0x7ff84f9bdc6d PyInit_pycolmap
@ 0x7ff84f99651d PyInit_pycolmap
@ 0x7ff84f954170 (unknown)
@ 0x7ff84f960a48 PyInit_pycolmap
@ 0x7ff84f96054e PyInit_pycolmap
@ 0x7ff856692304 _Py_fopen_obj
@ 0x7ff856692519 _Py_fopen_obj
@ 0x7ff85669246d _Py_fopen_obj
@ 0x7ff8565e09d7 _Py_HashPointer
@ 0x7ff8566193b4 PyVectorcall_Call