search

ml5js / ml5-library

Friendly machine learning for the web! 🤖
https://ml5js.org
Other
6.5k stars 900 forks source link

"Feature Extractor Examples" from ml5.js website / TypeError: Failed to fetch #1483

Open hedlerhdm opened 7 months ago

hedlerhdm commented 7 months ago

p5.js version

1.9.1

What is your operating system?

Windows 11

Web browser and version

Chrome 123.0.6312.106

Actual Behavior

the FeatureExtractor Examples for the "p5 web editor" from the ml5.js website raise an error: TypeError: Failed to fetch.

Here are the Examples: FeatureExtractor_Image_Regression FeatureExtractor_Image_Classification

image

a couple of days ago, the examples worked perfecty fine.

davepagurek commented 7 months ago

This is the error in the console for me:

Cross-Origin Request Blocked: The Same Origin Policy disallows reading the remote resource at https://www.kaggle.com/models/google/mobilenet-v1/frameworks/tfJs/variations/025-224-classification/versions/1/model.json?tfjs-format=file&tfhub-redirect=true. (Reason: CORS header ‘Access-Control-Allow-Origin’ missing). Status code: 302.

Looks like Kaggle recently made some API changes on April 8, so maybe it's related to that?

hedlerhdm commented 7 months ago

Besides the "Feature Extractor Examples", other sample codes are also affected: FaceMesh, Hand Pose, Image Classification, etc.

jht9629-nyu commented 7 months ago

I see cross-origin error on these two p5js sketches. They worked fine a few weeks ago.

https://editor.p5js.org/ml5/sketches/Facemesh_Webcam

https://editor.p5js.org/ml5/sketches/Handpose_Webcam

Access to fetch at 'https://www.kaggle.com/models/mediapipe/facemesh/frameworks/tfJs/variations/default/versions/1/model.json?tfjs-format=file&tfhub-redirect=true' (redirected from 'https://tfhub.dev/mediapipe/tfjs-model/facemesh/1/default/1/model.json?tfjs-format=file') from origin 'https://preview.p5js.org' has been blocked by CORS policy: No 'Access-Control-Allow-Origin' header is present on the requested resource. If an opaque response serves your needs, set the request's mode to 'no-cors' to fetch the resource with CORS disabled.

shiffman commented 7 months ago

Hi everyone! We are working on a new version of the ml5.js with more up-to-date models, the release is coming very soon! Please see the following for newer examples that you can use right now in the meantime!

New examples: https://editor.p5js.org/ml5/collections/pUzWMkdmE New library: https://unpkg.com/ml5@0.20.0-alpha.4/dist/ml5.min.js Temporary documentation: https://github.com/ml5js/ml5-next-gen/blob/main/documentation.md

Please file any issues in this repo: https://github.com/ml5js/ml5-next-gen

hedlerhdm commented 7 months ago

Thanks, Daniel! Your work is highly appreciated. Any chance to fix the problem anyway? I'm using ml5.js during ongoing lectures. Almost all examples are affected (e.g., feature extraction).

shiffman commented 7 months ago

Was this perhaps a temporary issue yesterday? Everything seems to be working for me, I tested the ones above as well as these three examples from my videos.

https://editor.p5js.org/codingtrain/sketches/qFZF7iDe https://editor.p5js.org/codingtrain/sketches/5A_TJHA1 https://editor.p5js.org/codingtrain/sketches/ULA97pJXR

jht9629-nyu commented 7 months ago

Working for me ok today.

These sketches now run OK: https://editor.p5js.org/ml5/sketches/Facemesh_Webcam https://editor.p5js.org/ml5/sketches/Handpose_Webcam

hedlerhdm commented 7 months ago

all examples are now working again. Maybe a temporary problem with Kaggle. Thanks anyway

devihall commented 7 months ago

Hey there! I can confirm this was a TensorFlow issue. I had raised this issue on their Github and they rolled out a fix. . They also provided some recommendations for how to avoid this in the future. See link attached.

https://github.com/tensorflow/hub/issues/904#issuecomment-2046088259

jht9629-nyu commented 3 months ago

Similar failure today:

https://docs.ml5js.org/#/reference/bodypose Access to fetch at 'https://www.kaggle.com/models/google/movenet/tfJs/multipose-lightning/1/model.json?tfjs-format=file&tfhub-redirect=true' (redirected from 'https://tfhub.dev/google/tfjs-model/movenet/multipose/lightning/1/model.json?tfjs-format=file') from origin 'https://docs.ml5js.org' has been blocked by CORS policy: No 'Access-Control-Allow-Origin' header is present on the requested resource. If an opaque response serves your needs, set the request's mode to 'no-cors' to fetch the resource with CORS disabled.

https://docs.ml5js.org/#/reference/facemesh Access to fetch at 'https://www.kaggle.com/models/mediapipe/face-landmarks-detection/tfJs/face-mesh/1/model.json?tfjs-format=file&tfhub-redirect=true' (redirected from 'https://tfhub.dev/mediapipe/tfjs-model/face_landmarks_detection/face_mesh/1/model.json?tfjs-format=file') from origin 'https://docs.ml5js.org' has been blocked by CORS policy: No 'Access-Control-Allow-Origin' header is present on the requested resource. If an opaque response serves your needs, set the request's mode to 'no-cors' to fetch the resource with CORS disabled.

shiffman commented 3 months ago

I just checked and it's working for me right now? Did it resolve for you @jht9629-nyu?

I think, unfortunately, we are at the mercy of the google/kaggle/tfhub servers and if something is down, then loading the model in ml5.js breaks. Maybe it would be good to document this better and/or have a back-up solution?

Feel free to file an issue in ml5-next-gen, that's where we are tracking ongoing development now!

https://github.com/ml5js/ml5-next-gen/

jht9629-nyu commented 3 months ago

Working for me now. Appears to be a temporary problem.

https://docs.ml5js.org/#/reference/bodypose https://docs.ml5js.org/#/reference/facemesh

jht9629-nyu commented 3 months ago

Since there is a hidden dependency on google servers to load models it would be great to have an example showing how to download those models and use them locally. ie. I would like to use facemesh without a web connection. Be well, -[ john-henry ]-[ thompson ]

On Tue, Aug 13, 2024 at 9:52 PM Daniel Shiffman @.***> wrote:

I just checked and it's working for me right now? Did it resolve for you @jht9629-nyu https://urldefense.proofpoint.com/v2/url?u=https-3A__github.com_jht9629-2Dnyu&d=DwMCaQ&c=slrrB7dE8n7gBJbeO0g-IQ&r=UCFvUQurfK8Qyy8EfTquZw&m=dm-rM_jSgyPlxtcthfNt2uKBSzCs3Lgd0lUrd9-4uE6NUB4r9AqKz3b-BapMWN1r&s=hEAQEgMkSRxbJ3J6uTv8JNhExu9mdtMgaZFlfAtY7eg&e= ?

I think, unfortunately, we are at the mercy of the google/kaggle/tfhub servers and if something is down, then loading the model in ml5.js breaks. Maybe it would be good to document this better and/or have a back-up solution?

Feel free to file an issue in ml5-next-gen, that's where we are tracking ongoing development now!

https://github.com/ml5js/ml5-next-gen/ https://urldefense.proofpoint.com/v2/url?u=https-3A__github.com_ml5js_ml5-2Dnext-2Dgen_&d=DwMCaQ&c=slrrB7dE8n7gBJbeO0g-IQ&r=UCFvUQurfK8Qyy8EfTquZw&m=dm-rM_jSgyPlxtcthfNt2uKBSzCs3Lgd0lUrd9-4uE6NUB4r9AqKz3b-BapMWN1r&s=MZ8ixm-LwETg1D7wBDpl0ZgHNpdMV1ecjUQ2vRH22HY&e=

— Reply to this email directly, view it on GitHub https://urldefense.proofpoint.com/v2/url?u=https-3A__github.com_ml5js_ml5-2Dlibrary_issues_1483-23issuecomment-2D2287681798&d=DwMCaQ&c=slrrB7dE8n7gBJbeO0g-IQ&r=UCFvUQurfK8Qyy8EfTquZw&m=dm-rM_jSgyPlxtcthfNt2uKBSzCs3Lgd0lUrd9-4uE6NUB4r9AqKz3b-BapMWN1r&s=viJ2HDrwtQpqpYtiJsEd0Gdc_dYomr5hraUMcoaDU_8&e=, or unsubscribe https://urldefense.proofpoint.com/v2/url?u=https-3A__github.com_notifications_unsubscribe-2Dauth_ASZWD3KELP5EOI3WY7JA25DZRKZ5RAVCNFSM6AAAAABF6T7W2WVHI2DSMVQWIX3LMV43OSLTON2WKQ3PNVWWK3TUHMZDEOBXGY4DCNZZHA&d=DwMCaQ&c=slrrB7dE8n7gBJbeO0g-IQ&r=UCFvUQurfK8Qyy8EfTquZw&m=dm-rM_jSgyPlxtcthfNt2uKBSzCs3Lgd0lUrd9-4uE6NUB4r9AqKz3b-BapMWN1r&s=K0TSj9iwD6GRgSp71YgQ2jsYURvNqX4e2fe7HPuxbDg&e= . You are receiving this because you were mentioned.Message ID: @.***>