ValueError: xyxy must be a 2D np.ndarray with shape (_, 4), but got shape (0, 0, 4)

[FrankTseng168]

Search before asking

• I have searched the Roboflow Notebooks issues and found no similar bug report.

Notebook name

how_to_finetune_florence_2_on_detection_dataset

Bug

Environment

win 11/ python3.11.10/ spyder

Minimal Reproducible Example

No response

Additional

No response

Are you willing to submit a PR?

• Yes I'd like to help by submitting a PR!

[LinasKo]

Hi @FrankTseng168 👋

Thank you for the report! I'm failing to reproduce it so far. Do you know when this happens?

Any chance you could print out the shape of prediction.xyxy, before prediction = prediction[np.isin(...)]? Also, are there any matches in classes, for prediction["class_name"]?

[FrankTseng168]

Hi, LinasKo ***@***.***> Thank you for your email。Please refer to the following result: prediction = sv.Detections.from_lmm(sv.LMM.FLORENCE_2, prediction, resolution_wh=image.size) print(prediction.xyxy) print(prediction.xyxy.shape) print(prediction['class_name']) prediction = prediction[np.isin(prediction['class_name'], CLASSES)] -> [[3.2000e-01 3.2000e-01 6.3904e+02 6.3904e+02]] (1, 4) ['bed'] [[3.2000e-01 3.2000e-01 6.3904e+02 6.3904e+02]] (1, 4) ['table'] [[3.2000e-01 3.2000e-01 6.3904e+02 6.3904e+02]] (1, 4) ['chair'] [[3.2000e-01 3.2000e-01 6.3904e+02 6.3904e+02]] (1, 4) ['furniture'] [[3.2000e-01 3.2000e-01 6.3904e+02 6.3904e+02]] (1, 4) ['bed'] [[3.2000e-01 3.2000e-01 6.3904e+02 6.3904e+02]] (1, 4) ['tablecloth'] [[3.2000e-01 3.2000e-01 6.3904e+02 6.3904e+02]] (1, 4) ['table'] [[198.72 182.08 227.51999 224.95999] [311.36 189.12 409.28 408.63998]] (2, 4) ['human face' 'person'] [[3.2000e-01 3.2000e-01 6.3904e+02 6.3904e+02]] (1, 4) ['bed'] [[3.2000e-01 3.2000e-01 6.3904e+02 6.3904e+02]] (1, 4) ['human face'] [] (0, 4) None LinasKo ***@***.***> 於 2024年11月27日 週三 下午10:38寫道：

…

Hi @FrankTseng168 <https://github.com/FrankTseng168> 👋 Thank you for the report! I'm failing to reproduce it so far. Do you know when this happens? Any chance you could print out the shape of prediction.xyxy, before prediction = prediction[np.isin(...)]? Also, are there any matches in classes, for prediction["class_name"]? — Reply to this email directly, view it on GitHub <#342 (comment)>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/AJ6ULVQWNFHXWDGERVZ3RXL2CXKORAVCNFSM6AAAAABSSNERTOVHI2DSMVQWIX3LMV43OSLTON2WKQ3PNVWWK3TUHMZDKMBUGA2DGOBQHA> . You are receiving this because you were mentioned.Message ID: ***@***.***>

[LinasKo]

Bug confirmed! Very peculiar. I'll let you know when I fix it.
Logged here: roboflow/supervision#1694

For now, try this:

if not predictions.is_empty():
    prediction = prediction[np.isin(prediction["class_name"], CLASSES)]

If it still fails, change to if len(predictions.xyxy) > 0:

[LinasKo]

The fix is now deployed to our develop branch. You can install it with pip install git+https://github.com/roboflow/supervision.git@develop.

Thank you for the report, @FrankTseng168!