Yolo26 Segmentation - Bad Inference Result

[ZachCostello]

Search before asking

• I checked the docs, issues, and discussions and could not find an answer.

Question

I recently trained a Yolo26 Segmentation model and am getting some crazy output in app. Below is my code in flutter and 3 images. I had this code running a Yolov11 model with no issues, then I just swapped the model for my new Yolo26 one and it seems to not be inferring properly. The Flutter Pub page for the plugin didn't indicate any changes that need to be made for Yolo26. When I run the the model in python on my computer, I get the very nicely detected image below. When I run it in my Flutter app, I get the unknown boxes with confidence values well over 100%. The image from my app is the 'annotedImage' that is part of the predict results (segmentResults['annotatedImage']).

Flutter Version: 0.2.0
Running on Android (physical device, Samsung S23)

try {
      //Current model (yolo26s-seg.pt) trained at imgsz=960 on 15 Jan 2026
      if (Platform.isIOS) {
        yolo = YOLO(
          modelPath: 'model.mlmodel',
          task: YOLOTask.segment,
          numItemsThreshold: 50,
        );
      } else {
        yolo = YOLO(
          modelPath: 'best_float32.tflite',
          task: YOLOTask.segment,
          numItemsThreshold: 50,
        );
      }

      await yolo!.loadModel();

      try {
      segmentResults = await yolo!.predict(widget.image);

Results from python (in Google Collab)

Results from Flutter app on Android

Image Tested On

Additional

No response

[UltralyticsAssistant]

👋 Hello @ZachCostello, thank you for submitting a ultralytics/yolo-flutter-app 🚀 Issue. This is an automated response to help us triage faster — an Ultralytics engineer will also assist soon 🛠️🙂. To help us address your concern efficiently (especially since this looks like a potential YOLO26 segmentation export/inference mismatch), please ensure you've provided the following information:

1. For bug reports:

• A clear and concise description of the bug (e.g., “YOLO26 segmentation shows ‘unknown’ boxes + confidence > 100% in Flutter Android, but works in Python”)
• A minimum reproducible example MRE that demonstrates the issue (please include a tiny sample image and the exact model artifact you’re loading in Flutter, if possible)
• Your environment details (OS, Python version, package versions)
• Expected behavior vs. actual behavior
• Any error messages or logs related to the issue (Android logcat output around model load + predict would be especially helpful 📋)

2. For feature requests:

• A clear and concise description of the proposed feature
• The problem this feature would solve
• Any alternative solutions you've considered

3. For questions:

• Provide as much context as possible about your question
• Include any research you've already done on the topic
• Specify which parts of the documentation, if any, you've already consulted

Please make sure you've searched existing issues to avoid duplicates. If you need to add any additional information, please comment on this issue.

Thank you for your contribution to improving our project! 🙏✨

[pderrenger]

This looks like a TFLite postprocess parsing mismatch (the “>100% confidence” is a classic symptom): YOLO26-seg exports return two output tensors and the tensor shapes/num_predictions depend on the export imgsz, so if you exported at 960 but the Flutter side assumes a different layout you’ll decode garbage boxes/scores; please try with the latest Flutter package and re-export your model to TFLite at imgsz=640, and if it still happens paste your exact yolo export ... command plus the two TFLite output tensor shapes so we can align the parser (see the Export docs FAQ on segmentation output tensors for the expected structure). (docs.ultralytics.com)

[asabri97]

@ZachCostello YOLO26 models don't need nms, you need to adjust your code to process yolo26 models. Same code won't run both models.

[github-actions]

👋 Hello there! We wanted to give you a friendly reminder that this issue has not had any recent activity and may be closed soon, but don't worry - you can always reopen it if needed. If you still have any questions or concerns, please feel free to let us know how we can help.

For additional resources and information, please see the links below:

• Docs: https://docs.ultralytics.com
• HUB: https://hub.ultralytics.com
• Community: https://community.ultralytics.com

Feel free to inform us of any other issues you discover or feature requests that come to mind in the future. Pull Requests (PRs) are also always welcomed!

Thank you for your contributions to YOLO 🚀 and Vision AI ⭐