MagicSource

icon indicating an email [email protected]

icon company Google icon location Sanfancisco icon location Play with Neural Magic, AIGC, 3D Computer Vision, Virtual Human, 3D Artist. work at @tencent

Results 1144 comments of MagicSource

训练问题

MyDatasetMapper没有做归一化，preprocees主要是做归一化。 第二点我建议你放弃。

WeChat number of group members has reached the limit

Hi, welcome join our discord channel: https://discord.gg/2E8AfjaEXM will push new wechat group invitation QRCode in channel.

Demo didnot save the output

the output should handled with wandadb now, check the code for logic.

How to perform inference for instance segmentation using the pre-trained model?

using sparsinst for instance seg, this is only one have pertrained weights.

Demo Trouble Shooting

I think you haven't sent weights. following the readme instruction provide a weight path.

Coco keypoints detection training

I don't know, I am gonna delete this repo soon.

Coco keypoints detection training

No, as for the official YOLOv7 released, I will make a place for them....

[t["labels"][J] for t, (_, J) in zip(targets, indices)] IndexError: tensors used as indices must be long, byte or bool tensors

Using `train_inseg.py` to train instance segmentation!

[t["labels"][J] for t, (_, J) in zip(targets, indices)] IndexError: tensors used as indices must be long, byte or bool tensors

不清楚，社区之前有人训过sparseinst，理论上应该是可以的。

Is that possible to do keypoint and Segmentation same time without effecting performance? nd get

Yes, try adding a segmentation head alongside YOLO detect head.