asset does not exist about ultralytics

[Laughing-q]

@AyushExel

Error executing job with overrides: ['task=detect', 'mode=predict', 'model=yolov8s.pt', 'source=0']

I think it's because hydra parsed the 0 in source=0 to str.

[AyushExel]

@Laughing-q yeah its treating 0 as null and predicts on the default assets. But it doesn't throw any error for me

[glenn-jocher]

@udayzee05 your issue may be resolved now in PR #256 that I just merged. Please update your code and let us know if this resolves the issue for you, thank you!

[AyushExel]

@udayzee05 thanks for reporting. Can you please provide the command to reproduce?

[udayzee05]

yolo task=detect mode=predict model=yolov8n.pt source="https://ultralytics.com/images/bus.jpg" I tried with this command as well but same result

[udayzee05]

Thank you, I was trying to pass 0 for webcam, but I am able to run successfully for images stored in local directory, Thank for clarification.

[AyushExel]

@udayzee05 Does this work for you now?

[udayzee05]

Yes, but for webcam what will be argument ?

[AyushExel]

Awesome

[Laughing-q]

@AyushExel

ultralytics/ultralytics/yolo/v8/detect/predict.py

Line 90 in 8302dc4

cfg.source = cfg.source or ROOT / "assets"

source becomes ultralytics/assets after this line if you pass source=0

[AyushExel]

@Laughing-q yeah that's what I meant.. hydra is treating "0" as None..not a string..we'll need a different syntax for webcam

[udayzee05]

Yes, even I tried to pass it as string as well as integer but getting error

[AyushExel]

yeah, I can reproduce. that is strange.

[zldrobit]

I could reproduce this problem on Colab with the default source argument, as follows

The version of ultralytics pip package is 8.0.6

[Laughing-q]

@zldrobit hey I tested with the same command by using this colab https://colab.research.google.com/github/ultralytics/ultralytics/blob/main/examples/tutorial.ipynb. But it works fine to me, probably there's something wrong with your installation, could check it again?

[zldrobit]

@Laughing-q I think I may find the origin of the problem. If I skip cloning the ultralytics GitHub repo, the error still occurs

Otherwise, I could also get detection results.

Plz note that the missing images are under /usr/local/lib/python3.8/dist-package, but the existing images are in /content/ultralytics being the ROOT path. I guess these images are not packaged in the PyPI wheel, so barely pip install is not able to download them.

[TIBREWA]

Hey @udayzee05 did you fix the ultralytics issue because i am facing the same issue and i am unable to fix it, if yes please tell me how can i fix it

[shaowngp]

My issue has been resolved after putting source= 'ultralytics/' ; providing Ultralytics root path.

!python predict.py model='best.pt' source='/content/gdrive/MyDrive/yolov8/ultralytics/' data='/content/gdrive/MyDrive/yolov8/ultralytics/xyz.mp4'

2023-11-03 05:49:40.932793: E tensorflow/compiler/xla/stream_executor/cuda/cuda_dnn.cc:9342] Unable to register cuDNN factory: Attempting to register factory for plugin cuDNN when one has already been registered
2023-11-03 05:49:40.932848: E tensorflow/compiler/xla/stream_executor/cuda/cuda_fft.cc:609] Unable to register cuFFT factory: Attempting to register factory for plugin cuFFT when one has already been registered
2023-11-03 05:49:40.932883: E tensorflow/compiler/xla/stream_executor/cuda/cuda_blas.cc:1518] Unable to register cuBLAS factory: Attempting to register factory for plugin cuBLAS when one has already been registered
2023-11-03 05:49:40.940297: I tensorflow/core/platform/cpu_feature_guard.cc:182] This TensorFlow binary is optimized to use available CPU instructions in performance-critical operations.
To enable the following instructions: AVX2 AVX512F FMA, in other operations, rebuild TensorFlow with the appropriate compiler flags.
2023-11-03 05:49:42.094048: W tensorflow/compiler/tf2tensorrt/utils/py_utils.cc:38] TF-TRT Warning: Could not find TensorRT
Ultralytics YOLOv8.0.3 🚀 Python-3.10.12 torch-2.1.0+cu118 CUDA:0 (Tesla T4, 15102MiB)
Fusing layers...
Model summary: 218 layers, 25840339 parameters, 0 gradients, 78.7 GFLOPs
image 1/2 /content/gdrive/MyDrive/yolov8/ultralytics/figure1.png: 384x640 1 licence, 74.4ms
image 2/2 /content/gdrive/MyDrive/yolov8/ultralytics/figure3.png: 384x640 1 licence, 25.8ms
Speed: 0.5ms pre-process, 50.1ms inference, 2.4ms postprocess per image at shape (1, 3, 640, 640)
Results saved to /content/gdrive/MyDrive/yolov8/runs/detect/train6

[glenn-jocher]

@shaowngp great to hear that your issue has been resolved by providing the correct root path for Ultralytics as your source. It appears the initially configured root path couldn't find the necessary files. Now, the program can successfully locate them and everything runs smoothly. The TensorFlow errors that are being shown are warnings and should not impact the overall functionality of YOLOv8, those are related to the TensorFlow-TensorRT integration and not finding TensorRT, which is a high-performance deep learning inference optimizer and runtime library. Lastly, Ultralytics YOLOv8 seems to be working efficiently as indicated by the speed of the pre-process, inference, and postprocess per image.