search

arXiv / html_feedback

Supports a student project developing a UI for feedback on arXiv articles rendered as html.
MIT License
17 stars 3 forks source link

无法翻译 #1950

Closed gottang closed 1 week ago

gottang commented 3 weeks ago

Description

无法翻译

(Optional:) Please add any files, screenshots, or other information here.

No response

(Required) What is this issue most closely related to? Select one.

Choose One

Internal issue ID

69cd0365-c3d7-4d43-894d-fcbce5a88341

Paper URL

https://arxiv.org/html/2403.09309?_immersive_translate_auto_translate=1

Browser

Chrome/99.0.4844.84

Device Type

Desktop

html-feedback-bot[bot] commented 3 weeks ago

Location in document: S3.SS1.p1.2

Selected HTML: 𝑁Nitalic_N learned embeddings called object queries to generate N𝑁Nitalic_N object embeddings, which are then processed by feed-forward prediction networks (FFPNs) to generate class probabilities, 2D bounding box, and IBB keypoints in parallel. The IBB keypoints are then processed by a subsequent FFPN to estimate the translation and rotation parameters. Since the cardinality of the predicted set is fixed, the model is trained to predict Ø classes after detecting all the target objects present in the image. By associating predictions and ground truth objects using a bipartite matching algorithm [46], YOLOPose is trained end-to-end.

github-actions[bot] commented 3 weeks ago

Hello @gottang, thanks for the issue report! We are reviewing your report and will address it as soon as possible.