human-pose-estimation.pytorch
human-pose-estimation.pytorch copied to clipboard
microsoft
The project is an official implement of our ECCV2018 paper "Simple Baselines for Human Pose Estimation and Tracking(https://arxiv.org/abs/1804.06208)"
https://github.com/microsoft/human-pose-estimation.pytorch/blob/master/lib/dataset/JointsDataset.py#L85 您好，请问下，代码中center， scale部分。 请问下： 1.这里的center是标注人的目标框的中心坐标吗？ 2.喂入网络模型的输入是整张图还是根据目标框裁剪好的图呢？ - 如果喂入的是整张图，那么center是整张图的中心， 还是目标框对应的中心呢？ - 如果是裁剪好的图，是否是目标框对应中心？
I am working on another project for which I need to understand the pose of the subject. I hope to use the pre-trained model given, but I can't load the...
I checked the MPII JSON annotations you posted in Google Drive, however, i found there're many annotations are missing from original MPII matlab version. Here're some examples: original MPII annotations:...
作者您好，关于数据增强，我想请教下。 https://github.com/microsoft/human-pose-estimation.pytorch/blob/master/experiments/coco/resnet50/384x288_d256x3_adam_lr1e-3.yaml#L14 配置文件中，`ROT_FACTOR: 40`、`SCALE_FACTOR: 0.3`。 然后数据集处理过程中 https://github.com/microsoft/human-pose-estimation.pytorch/blob/master/lib/dataset/JointsDataset.py#L93 `s = s * np.clip(np.random.randn()*sf + 1, 1 - sf, 1 + sf)` `r = np.clip(np.random.randn()*rf, -rf*2, rf*2) \ if random.random()
作者您好，关于数据增强，我想请教下。 https://github.com/microsoft/human-pose-estimation.pytorch/blob/master/experiments/coco/resnet50/384x288_d256x3_adam_lr1e-3.yaml#L14 配置文件中，`ROT_FACTOR: 40`、`SCALE_FACTOR: 0.3`。 然后数据集处理过程中 https://github.com/microsoft/human-pose-estimation.pytorch/blob/master/lib/dataset/JointsDataset.py#L93 `s = s * np.clip(np.random.randn()*sf + 1, 1 - sf, 1 + sf)` `r = np.clip(np.random.randn()*rf, -rf*2, rf*2) \ if random.random()
Hi, I am confusing about heatmap generation. 1. From COCO dataset, the valid of joint are 0 for not in the image, 1 for in the image but not visible,...
I experiment your framework with different backbone. Surprisingly, the performance of resnet34 is quite close to resnet50. Renest34 have mAP of about 0.705 (and 0.893 [email protected]) on 2017 coco keypoints...
Is there any mobile implementation? Androis / iOS? Thanks
@leoxiaobin first thank you for your job,but i try to use warpAffine & gaussian_filter & maximum_filter to draw the result ,show offset error,could you tell how to convert the hm...
