Evan

Wandb logging bug using Iteration based runner

12

**Describe the Issue** Validation metrics reporting/logging to wandb does not happen when using `IterBasedRunner` **Reproduction** Here's a simple reproduction of the bug. Using mmdetection, In config file, `faster_rcnn/faster_rcnn_r50_caffe_c4_1x_coco.py` with the...

File client for reading from S3 Object Storage

8

Hi! I am directed here [from mmdetection](https://github.com/open-mmlab/mmdetection/issues/6470). It seems like `petrel_client` is something internal to your lab. Are there plans to support a file client for S3 Object Storage?

Log config to wandb

6

Although there is `WandbLoggerHook`, but the config of the run is not logged to wandb. Will be nice if this can be done by default as long as `WandbLoggerHook` is...

Trained model weights for "Strong Baselines"

1

Hi, will like to check where can we find the trained model weights for [strong baselines](https://github.com/open-mmlab/mmdetection/tree/master/configs/strong_baselines)? The links to all the "model" does not point to the weights download url.

Multi-Instance Pose Networks (MIPNet)

2

Hi! Firstly, thank you for this wonderful library. May I know if there are any plans to implement [MIPNet](https://github.com/rawalkhirodkar/MIPNet) in mmpose? It is particularly useful to tackle data where there...

Hi, thank you for the nice work here. Can I understand how you obtained `person_keypoints_train2017_pose2seg.json` and `person_keypoints_val2017_pose2seg.json`? Was it obtained by filtering out the "person" class from the COCO dataset...

Source of images

1

Hi! Thank you for the dataset, it is very useful. Can I check where are the images sourced from? I can't find that info in your paper.

Using ConcatDataset in validation set in new mmdet v3

2

For the latest release mmdet v3, if `ConcatDataset` is used in validation dataset, how can `val_evaluator` be set in the config? I've tried not providing ann_file to `val_evaluator` as well...

https://github.com/open-mmlab/mmdetection/blob/c9b47820d637ab341ca3ca869116fc3d91030c74/mmdet/visualization/local_visualizer.py#L200 This line throws a ValueError when there are no predictions, as `areas` is empty. Easy fix will be to check for empty predictions.

cache_randomness explanation

3

Hi @ZwwWayne, saw that the latest release uses quite a bit of `cache_randomness` in the augmentation pipelines. May I know what is the function of these decorators and how it...