Jianwei Yang

I will figure out that. thanks for letting me know.

@ParsonsZeng thanks for reminding, I will update the links.

> Florence https://arxiv.org/abs/2111.11432 Please refer to our UniCL repo on the core algorithm used in Florence: https://github.com/microsoft/UniCL

Hi, all, If possible, it would be good to share your data loaders with us, so that we can check why that happened.

Hi, @sdlyaoyuan , I will take a look at this problem, and try to solve it as soon as possible.

@1160300929 the number of GPUs I used is listed in the tables. I trained with 8 gpus, and the time for training scene graph generation model given the pertained detector...

Hi, both, I am still working on it to figure out why the performance is worse than faster rcnn. It seems that I found a small bug in my code...

@cjrd , thanks for reminding, I am going to upload the config files for each model.

Hi, thanks for your post. The off-the-shelf model is not applicable to rencent code any longer, sorry that i did not stated that difference. 1200 or 300 mentioned in the...

你好，你的做法没有问题，你可以把face框出来之后，然后用一个平均脸作为你的groundtruth或者任意初始化生成一个grountruth pt文件即可。 > On Dec 7, 2017, at 8:34 PM, fuzixiansheng wrote: > > 非常好的分享！感谢！ > 我成功地在你们提供的图片库中运行成功了，结果很好。 > 但是目前我想用自己的图片试试，而该图片缺少.pts文件，参考了之前的提问，我选择用matlab自带的工具箱来圈出脸部边界。至于landmarks(也就是shape_gt变量)，我是随机赋值的。结果发现shape_gt没有被更新，也就是没有程序没有找到landmarks，想请教一下这是什么原因呢，该怎么解决呀？ > 我的想法是shape_gt这个变量是标记点的坐标，应该会被模型重新打上标记，所以我随机初始化了，这是为了匹配上程序的输入数据的格式，不知道是不是我这样的想法有问题。还望赐教！ > > — > You are receiving this because...