-
### What is the feature?
Would it be possible to have a list of all possible checkpoints available for the backbones?
For e.g topdown heat map for hrnetw48_udp uses hrnet as a backbone. However t…
-
I want to convert HRNET model for pose estimation:
+) checkpoint hrnet_w32_coco_256x192-c78dce93_20200708.pth
+) config: hrnet_w32_coco_256x192.py.
I try to use command in mmaction2/tools/deployme…
-
How can I get the results in the label from the results that came out on test.py?
python tools/test.py configs/hrnet/hrnet_config.py checkpoints/iter_40000.pth --show-dir result_test
-
Hello, I used your seg_hrnet_w18_small_v2 model for training, and I also used the pre-trained model you gave. But the obtained training mIoU is only 0.69.Excuse me, where should I modify it?
-
请问在w48_384x288_adam_lr1e-3.yaml配置文件中有
DATASET::
LOCREF_STDEV: 1.5
和
LOSS:
OFFSET_WEIGHT: 1.0
然后在其他分辨率的配置文件就没有 请问是不需要这个设置了吗
-
Thanks for sharing!
Looking around this repo, I found that the training code is made only for `pose_hrnet`. This is because `pose_resnet` only return heatmap, not including offsetmap. Similarly, hr…
-
Are you sure that this method is faster than HRNet?
I have tried both with yolov5 as the detector in trt inference.
HRNet achieves around 30-35 fps while VitPose can reach 7 fps at the same video wi…
-
Traceback (most recent call last):
File "tools/train.py", line 248, in
main()
File "tools/train.py", line 137, in main
normalize,
File "E:\DeepFashion2_Dataset\HRNet-Human-Pose-Est…
-
paddlepaddle-gpu==0.0.0.post110
paddlehub==2.1.1
以下方式都失败了:
方式1:
import paddlehub as hub
human_seg = hub.Module(name="FCN_HRNet_W18_Face_Seg")
human_seg.save_inference_model(dirname="output/"…
-
i downloaded the hrnet18 model from this repository and using the following command i can train it without the 'resume' flag
```python
python train.py mot --exp_id all_hrnet --gpus '0' --batch_size …