search

ZJieX / prsnet

PRSNet: A Masked Self-Supervised Self-Supervised Learning Pedestrian Re-Identification Method
3 stars 0 forks source link

Question regarding evaluation on MSMT17 #3

Open kennyvoo opened 1 year ago

kennyvoo commented 1 year ago
Hi. May I know if the result of the MSMT17 is with or without reranking? and if it is computed based on USE_CENTROIDS=True? mAP rank1
86.4 83.4
ZJieX commented 1 year ago

Thank you for your interest in the job. No ReRanking was used，and the centroid was used during training, but not during testing.

kennyvoo commented 1 year ago

do you plan to release the pretrained weight?

ZJieX commented 1 year ago

I'm very sorry, because I'm busy with work recently, so I didn't come to reply urgently, because the pre-trained models are all obtained through self-supervised learning, and then used for training in pedestrian re-identification. The whole training process only takes less than one night. Yes, so I recommend training by yourself. If you need the model I trained, please send it to your online disk, and I will send it to you.

kennyvoo commented 1 year ago

Do you mind sharing the pretrained weight ? https://drive.google.com/drive/folders/1el795mzjraQ-LJJWLUN7hFyaK9OgE0if?usp=sharing

ZJieX commented 1 year ago

I'm really sorry and thank you very much for your interest in this project. Due to being too busy with work, I haven't logged into this email account. I will share the trained model from my side in a few days. I also recommend that you can follow the steps in the README.md file to obtain the final results by yourself. If you have any questions, please feel free to contact me. I wish you all the best in life! Thanks！

---- Replied Message ---- | From | Kenny @.> | | Date | 11/23/2023 16:46 | | To | ZJieX/prsnet @.> | | Cc | xzj @.>, Comment @.> | | Subject | Re: [ZJieX/prsnet] Question regarding evaluation on MSMT17 (Issue #3) |

Do you mind sharing the pretrained weight ? https://drive.google.com/drive/folders/1el795mzjraQ-LJJWLUN7hFyaK9OgE0if?usp=sharing

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