search

AlibabaResearch / efficientteacher

A Supervised and Semi-Supervised Object Detection Library for YOLO Series
GNU General Public License v3.0
831 stars 151 forks source link

支持yolov8半监督训练吗 #11

Closed Bigtuo closed 1 year ago

BowieHsu commented 1 year ago

We have implemented a version that is still in testing. Essentially, it just involves adding an additional ssod loss for YOLOv8.

Bigtuo commented 1 year ago

那这个v8什么时候会放出来呢，还有你们这些算法，去实现分割功能好修改吗

---原始邮件--- 发件人: @.> 发送时间: 2023年3月3日(周五) 中午12:00 收件人: @.>; 抄送: @.**@.>; 主题: Re: [AlibabaResearch/efficientteacher] 支持yolov8半监督训练吗 (Issue #11)

We have implemented a version that is still in testing. Essentially, it just involves adding an additional ssod loss for YOLOv8.

— Reply to this email directly, view it on GitHub, or unsubscribe. You are receiving this because you authored the thread.Message ID: @.***>

BowieHsu commented 1 year ago

Hello, due to our limited development resources, I will provide a scheduling update for your two requirements. Currently, we are working on semi-supervised anchor-free detection, as you can see, we have already refactored the main YOLO series solutions into a unified format and are conducting extensive experiments. As for supporting segmentation, our refactoring solution is actually very similar to YOLOv8. You can modify the dataloader to read segmentation data and then reload a segmentation detection head to complete the segmentation training task (these are readily available implementations in YOLOv8, so it shouldn't be too difficult to modify) Feel free to ask any question!

Bigtuo commented 1 year ago

好的，谢谢😊

---原始邮件--- 发件人: @.> 发送时间: 2023年3月3日(周五) 中午12:24 收件人: @.>; 抄送: @.**@.>; 主题: Re: [AlibabaResearch/efficientteacher] 支持yolov8半监督训练吗 (Issue #11)

Hello, due to our limited development resources, I will provide a scheduling update for your two requirements. Currently, we are working on semi-supervised anchor-free detection, as you can see, we have already refactored the main YOLO series solutions into a unified format and are conducting extensive experiments. As for supporting segmentation, our refactoring solution is actually very similar to YOLOv8. You can modify the dataloader to read segmentation data and then reload a segmentation detection head to complete the segmentation training task (these are readily available implementations in YOLOv8, so it shouldn't be too difficult to modify) Feel free to ask any question!

— Reply to this email directly, view it on GitHub, or unsubscribe. You are receiving this because you authored the thread.Message ID: @.***>

jie311 commented 1 year ago

想问一下基于yolov8的ssod完成了吗