Unable to create recognition file - for dataset - 'local variable' 'elapsed_time' referenced before assignment

DRW · February 26, 2026, 10:21pm

Hello,

I have been using Addax AI through TimeLapse for about a year and this is the first error message I have received. I recently upgraded to version 6.29 and downloaded the Parks Victoria - Addax data science model. When I tried to run it for the first time it was “locating animals” for about 20 hours then I got an error message saying a recognition file could not be created referencing a local variable called “elapsed_time” which was referenced before assignment, see error message below.

Can someone please tell me what this message means?

petervanlunteren · February 27, 2026, 2:58pm

Hi @DRW ,

That error message is generic and means something went wrong during analysis. Does it happen more often? Or was this a one-time error? If it happens consitently, could you run it in debug mode and share the logs with me?

You can send it to peter@addaxdatascience.com

That will give me more information on whats going on.

Thanks!

Cheers,

Peter

petervanlunteren · February 27, 2026, 3:01pm

On second thought, it might just be a corrupted model download. Could you:

make sure AddaxAI is not running, and the app is closed.
remove this entire folder with contents: C:\Users\<username>\AddaxAI_files\models\cls\Victoria, Australia - Parks Victoria - Addax Data Science
Restart AddaxAI. It will now show you a window saying there is a new model available, the VIC model
Try running the model on some data and click yes to download
make sure you have stable internet during the entire process. I believe the model is about 500MB.

Does that solve it?

Cheers,

Peter

chrisdieck · April 30, 2026, 1:40pm

I have a similar issue:

I have tried to delete and re-install the model in question and tried using a previous version of the Europe DeepFaune model, with no success.

My internet connection is unstable, which could affect the integrity of the download?

Can anyone help?

TIA

petervanlunteren · May 4, 2026, 11:04am

Hi @chrisdieck ,

Yes, if you have an unstable internet connection, that is almost cerntly the issue. Especially since DeepFaune is such a large model. There are instructions on how to donwload it manually below. Can you try that?

github.com/PetervanLunteren/AddaxAI

markdown/manual-download-weights.md

main


# Manually download model weights

Below are instructions for users who are unable to download model weight files through AddaxAI due to network restrictions (e.g., university or government networks). This should be plan B, as the automatic download is much quicker. If you haven't tried the automatic download yet, try that first and check your firewall, proxy, and VPN settings. If you still have trouble, feel free to reach out at [peter@addaxdatascience.com](mailto:peter@addaxdatascience.com).

Some models consist of a single file, others of several (model weights, class list, taxonomic mapping, sometimes a pretrained backbone). You need to download **all listed files** for a given model and place them together in the same folder.

## Step-by-step instructions

1. Find the model you need in the list below. The model name shown matches the name displayed in AddaxAI's error window.

2. Click each download link and save the file with the **exact filename** shown. Filenames must remain unchanged.

3. Navigate to your `AddaxAI_files` folder. The location depends on your operating system, see [this page](https://github.com/PetervanLunteren/AddaxAI/blob/main/markdown/AddaxAI_files_location.md) for details.

4. Open the target folder shown for your model inside `AddaxAI_files`. If the folder does not already exist, create it. The folder name must match exactly, including spaces, punctuation, and special characters (e.g., `Hawaiʻi, USA - AI Puaʻa v1.0`).

5. Move all downloaded files into that model folder. For example:
   ```
   AddaxAI_files/models/cls/Namibian Desert - Addax Data Science/namib_desert_v1.pt

This file has been truncated. show original

chrisdieck · May 5, 2026, 9:19am

Thanks for the suggestion. I have now manually downloaded the model and have placed it in the requisite folder.

I now receive this error:

petervanlunteren · May 7, 2026, 9:27am

Hi @chrisdieck ,

It seems like the variables.json is somehow corrupted. Can you:

open this file in a text editor like TextEdit, NotePad or VScode? C:/Users/chris/AddaxAI_files/models/cls/Europe - DeepFaune v1.4/variables.json
The contents are probabaly empty or partly missing. Make sure you have exactly the contents below.


{
    "model_fname": "deepfaune-vit_large_patch14_dinov2.lvd142m.v4.pt",
    "description": "The Deepfaune initiative aims at developing 'artificial intelligence' models to automatically classify species in images and videos collected using camera-traps. The initiative is led by a core academic team from the French 'Centre National de la Recherche Scientifique' (CNRS), in collaboration with more than 50 European partners involved in wildlife research, conservation and management. The Deepfaune models can be run through a custom software freely available on the website, or through other software packages or platforms like EcoAssist. New versions of the model are published regularly, increasing classification accuracy or adding new species to the list of species that can be recognized. More information is available at: https://www.deepfaune.cnrs.fr",
    "developer": "The DeepFaune initiative",
    "env": "pytorch",
    "type": "deepfaune-v1.4",
    "download_info": [
        [
            "https://huggingface.co/Addax-Data-Science/Deepfaune_v1.4/resolve/main/deepfaune-vit_large_patch14_dinov2.lvd142m.v4.pt?download=true",
            "deepfaune-vit_large_patch14_dinov2.lvd142m.v4.pt"
        ]
    ],
    "citation": "https://doi.org/10.1007/s10344-023-01742-7",
    "license": "https://creativecommons.org/licenses/by-sa/4.0/",
    "total_download_size": "1.1 GB",
    "info_url": "https://www.deepfaune.cnrs.fr/en/",
    "all_classes": [
        "bison",
        "badger",
        "ibex",
        "beaver",
        "red deer",
        "golden jackal",
        "chamois",
        "cat",
        "goat",
        "roe deer",
        "dog",
        "raccoon dog",
        "fallow deer",
        "squirrel",
        "moose",
        "equid",
        "genet",
        "wolverine",
        "hedgehog",
        "lagomorph",
        "wolf",
        "otter",
        "lynx",
        "marmot",
        "micromammal",
        "mouflon",
        "sheep",
        "mustelid",
        "bird",
        "bear",
        "porcupine",
        "nutria",
        "muskrat",
        "raccoon",
        "fox",
        "reindeer",
        "wild boar",
        "cow"
    ],
    "selected_classes": [
        "bison",
        "badger",
        "ibex",
        "beaver",
        "red deer",
        "golden jackal",
        "chamois",
        "cat",
        "goat",
        "roe deer",
        "dog",
        "raccoon dog",
        "fallow deer",
        "squirrel",
        "moose",
        "equid",
        "genet",
        "wolverine",
        "hedgehog",
        "lagomorph",
        "wolf",
        "otter",
        "lynx",
        "marmot",
        "micromammal",
        "mouflon",
        "sheep",
        "mustelid",
        "bird",
        "bear",
        "porcupine",
        "nutria",
        "muskrat",
        "raccoon",
        "fox",
        "reindeer",
        "wild boar",
        "cow"
    ],
    "var_cls_detec_thresh": "0.35",
    "var_cls_detec_thresh_default": "0.35",
    "var_cls_class_thresh": "0.80",
    "var_cls_class_thresh_default": "0.80",
    "var_smooth_cls_animal": true,
    "min_version": "6.20",
    "var_tax_levels_idx": 0,
    "var_tax_fallback": false,
    "var_tax_fallback_default": false
}

Then restart AddaxAI. Does it work?

angushouston · May 8, 2026, 3:39am

I also have a similar issue, using the parks Vic model and running it off CPU. A similar thing happened on Monday and i since deleted and reinstalled Addax. Thanks!!

petervanlunteren · May 11, 2026, 7:47am

Hi @angushouston ,

It looks like it is a corrupted download of the PT model file. Are you by any chance on an unstable internet conection? Then I would recomend to install the model file manually, as described here.

Make sure to remove the already existing models/cls/Victoria, Australia - Parks Victoria - Addax Data Science/ folder first, then replace by the manually downloaded files.

Let me know how that goes!

Cheers,

Peter

chrisdieck · May 11, 2026, 9:23am

Thanks again for the support. Sorry it takes me so long to try these suggestions.

The .json file was missing, so I pasted the exact text you posted into a blank NotePad file and saved it as variables.json and saved it to the DeepFaune folder.

Dunno if that was the right thing to do but I am now back to the original ‘local variable ‘elapsed_time. referenced before assignment’ error.

petervanlunteren · May 11, 2026, 11:51am

Hi @chrisdieck ,

No worries! We can delve into your bug and see if we can get it working, but in the meantime I’m also beta testing the new version. So it might be better to see if the new version fixes this for you, so we dont have to spend effort in a soon-to-be-depreciated version.

What do you think? You can give the new version a try here: New AddaxAI v7 ready for beta testing!

Cheers,

Peter

chrisdieck · May 11, 2026, 12:20pm

Hi @petervanlunteren

I don’t think it’s worth spending more time on this and I will give the new version a try.

Thanks

Chris

chrisdieck · May 14, 2026, 10:58am

I have resolved the original issue and it seems to have due to my unstable internet connection. With a stable connection, I have been able to get v6.3 to work with deepFaune 1.4.

I did have a go at using the v7 beta and was able to download and start that, again needed a stable connection, but was stumped when I could make no ore progress, presumably due to the files having no GPS or time metadata. Am now playing with v6.3 again.

Thanks for the help.

petervanlunteren · July 28, 2026, 9:07am

I just released AddaxAI v7. It’s completely rebuild from scratch. This issue should be resolved now. Could you test it for me and let me know?

Thanks!

Download: https://www.addaxai.com/
Documentation: https://docs.addaxai.com/

chrisdieck · August 10, 2026, 6:55am

Hi Peter,

Apologies, I have been in the field for the last few weeks and am only just catching up.

I will give it a go as soon as I have caught up with everything else.

Is there anything in particular you want me to keep an eye on?

Regarding the model only downloading successfully on a stable Internet connection: Did you want me to try and download it again or would I be able to copy the files i previously downloaded? The reason I ask is if I had to use a stable connection again, i would have to wait until the next time I am in the office, which could be another few weeks.

Chris

petervanlunteren · August 10, 2026, 1:51pm

Hi Chris,

I see. No, the models and the environments do not ship from the old version to the new. The reason is because the new one extaects them in a different format.

If redownloading them is difficult, then never mind! Not worth it Just keep using the version you have. If it works, it works.

Cheers,

Peter

chrisdieck · August 11, 2026, 1:45pm

Hi Peter,

have just run a test on >3k image files. My network held stable long enough to download the DeepFaune model.

Everything seemed to work just fine and the manual verification step is much improved and the outputs look very good. I’ve not had time to go through the results in detail but they look good at a quick glance.

I did have an issue with the timestamps. My files are video files cut split into frames, and as such, the file creation times do not match when the videos were recorded. As the filenames contain the date and time that is a small issue only.

Cheers

Chris