-
## Instructions To Reproduce the 🐛 Bug:
1. Full runnable code :
```
`#!/usr/bin/env python
# Copyright (c) Facebook, Inc. and its affiliates.
# this is an adaptation of detectron2/tools/deploy/ex…
-
Hi,
I trained different default model configurations (mask_rcnn, cascade) with different backbones (swin, r50, x101) for a custom COCO formatted dataset. The dataset contains only one class with u…
-
File "./train.py", line 194, in
main()
File "./train.py", line 184, in main
train_detector(model,
File "/home/ddh/anaconda3/envs/mm38_Detec_cu11_1/lib/python3.8/site-packages/mmdet/…
-
-
Hello, I want to train the masktrack_rcnn with the coco dataset. So i had reset the **dataset** of masktrack_rcnn_r50_fpn_12e_youtubevis2019.py-----`-'../../_base_/datasets/coco_instance.py'` and the …
-
您好!我在使用maskrcnn的config作用于wildreceipt数据集时,config文件如下
"""
####################################################################################################
# Copyright Info : Copyright (c) Dava…
-
Hi,
I'm trying to reproduce Your results in instance segmentation, using the scripts that you delivered (train_maskyolo_step1.sh and 2). I did everything according to the instruction. The scripts wor…
-
unexpected key in source state_dict: fc.weight, fc.bias
E:\mmdetection\week2_mmdet\Week2_mmdet\mmdetection-2.11.0\mmdet\datasets\custom.py:154: UserWarning: CustomDataset does not support filtering…
-
In my parameter Settings, I annotated all the MASK part of HTC, at this time I saw that the program can be trained, and the RPN part used HTC module. Is this using the HTC module?
-
My server does not have 8 GPUs. When I use 4 GPUs for training (without any modification to the network), I just change num-gpus to 4. After training, the segmentation accuracy is always 0.00. Do you …