Hi everyone! We are beyond excited to announce the addition of weakly-supervised segmentation labels for the WCS train and test data. The labels come from the DeepMAC class-agnostic segmentation model, details about the format can be found on github.
\nWe've added a kernel that demonstrates how to load and visualize the segmentations here.
", "messages": [ { "id": "1261003", "postDate": "04/02/2021 15:48:59", "content": "Hi everyone! We are beyond excited to announce the addition of weakly-supervised segmentation labels for the WCS train and test data. The labels come from the DeepMAC class-agnostic segmentation model, details about the format can be found on github.
\nWe've added a kernel that demonstrates how to load and visualize the segmentations here.
", "rawMarkdown": "Hi everyone! We are beyond excited to announce the addition of weakly-supervised segmentation labels for the WCS train and test data. The labels come from the [DeepMAC](https://google.github.io/deepmac/) class-agnostic segmentation model, details about the format can be found on [github](https://github.com/visipedia/iwildcam_comp/blob/master/readme.md#class-agnostic-segmentation-model).\n\nWe've added a kernel that demonstrates how to load and visualize the segmentations [here](https://www.kaggle.com/vighneshbgoogle/iwildcam-visualize-instance-masks).", "votes": null }, { "id": "1268614", "postDate": "04/09/2021 15:29:05", "content": "We've just added additional code that demonstrates how to run the DeepMac model on iWildCam data with your own boxes here, in case you want to try with MegaDetectorV4 or your own detector!
", "rawMarkdown": "We've just added additional code that demonstrates how to run the DeepMac model on iWildCam data with your own boxes [here](https://www.kaggle.com/vighneshbgoogle/run-deepmac-inference-for-iwildcam), in case you want to try with MegaDetectorV4 or your own detector!", "votes": null } ], "comments": [ { "id": 1268614, "author_name": "sbeery", "author_url": "", "post_date": "04/09/2021 15:29:05", "content": "We've just added additional code that demonstrates how to run the DeepMac model on iWildCam data with your own boxes here, in case you want to try with MegaDetectorV4 or your own detector!
", "votes": null, "replies": [] } ], "raw_markdown_by_id": { "1261003": "Hi everyone! We are beyond excited to announce the addition of weakly-supervised segmentation labels for the WCS train and test data. The labels come from the [DeepMAC](https://google.github.io/deepmac/) class-agnostic segmentation model, details about the format can be found on [github](https://github.com/visipedia/iwildcam_comp/blob/master/readme.md#class-agnostic-segmentation-model).\n\nWe've added a kernel that demonstrates how to load and visualize the segmentations [here](https://www.kaggle.com/vighneshbgoogle/iwildcam-visualize-instance-masks).", "1268614": "We've just added additional code that demonstrates how to run the DeepMac model on iWildCam data with your own boxes [here](https://www.kaggle.com/vighneshbgoogle/run-deepmac-inference-for-iwildcam), in case you want to try with MegaDetectorV4 or your own detector!" }, "source": "meta" }