YOLOX with onnxruntime-web
hello, do you have any examples or tutorials of running YOLOX with onnxruntime-web
I have YOLOX model working successfully, passing in a tensor and getting an output tensor. But I am struggling a bit to convert the output tensor to predictions and bounding boxes
https://github.com/Megvii-BaseDetection/YOLOX/blob/237e943ac64aa32eb32f875faa93ebb18512d41d/yolox/utils/demo_utils.py#L99
Sorry I could not catch your point. Could you say it a bit clear? Anything wrong in demo_postprocess?
sorry. I am trying to run YOLOX on a web app. So all the code is in javascript. I can use onnx-runtime web to get the output.
but without numpy in javascript I dont fully understand how to process the output to get the predictions.
so I was wondering if there is an example of anyone running YOLOX in a web / javascript.
Sorry we never try such a web application. What a about writing JS code to handle the post processing process and send a PR to this repo? @pconway11
@pconway11 @FateScript Hey have you found any example of usage on the web application? I tried to run this model on this repos:
- https://github.com/Hyuto/yolov8-tfjs
- https://github.com/Hyuto/yolov5-tfjs But unfortunately, it didn't work in any case.
I basically transcribed the python example into javascript. I didn’t find any example on the web.
It mostly worked, but I didn't implement NMS. In many cases in my experiments there was only one bounding box detection in the results so I didn't need NMS.
Sent from Outlook for iOShttps://aka.ms/o0ukef
From: Filip Głowicki @.> Sent: Wednesday, January 29, 2025 8:49:36 PM To: Megvii-BaseDetection/YOLOX @.> Cc: Patrick Conway @.>; Mention @.> Subject: Re: [Megvii-BaseDetection/YOLOX] YOLOX with onnxruntime-web (Issue #1111)
CAUTION: This email originated from outside of the organization. Do not click links or open attachments unless you recognize the sender and know the content is safe.
@pconway11https://github.com/pconway11 @FateScripthttps://github.com/FateScript Hey have you found any example of usage on the web application? I tried to run this model on this repos:
- https://github.com/Hyuto/yolov8-tfjs
- https://github.com/Hyuto/yolov5-tfjs But unfortunately, it didn't work in any case.
— Reply to this email directly, view it on GitHubhttps://github.com/Megvii-BaseDetection/YOLOX/issues/1111#issuecomment-2622816714, or unsubscribehttps://github.com/notifications/unsubscribe-auth/ADC2XU2ZQQAHW2WT53LNC7T2NE5GBAVCNFSM6AAAAABWDYJFR2VHI2DSMVQWIX3LMV43OSLTON2WKQ3PNVWWK3TUHMZDMMRSHAYTMNZRGQ. You are receiving this because you were mentioned.Message ID: @.***>