{ "id": 214925, "title": "Summary of Human Protein Atlas - 1st Competition (Jan 2019)", "url": "/competitions/hpa-single-cell-image-classification/discussion/214925", "author_name": "SkyLord", "post_date": "2021-01-28T05:51:42.152000", "votes": 3, "comment_count": 0, "views": 0, "content": "

The HPA-1 was held between Dec/Nov 2018-Jan2019. The objective was to classify \"mixed protein cells\" Historically, classification of proteins has been limited to single patterns in one or a few cell types, but in order to fully understand the complexity of the human cell, models must classify mixed patterns across a range of different human cells.

\n

The main difference between the 1st & 2nd HPA competition, is that in the current version (2nd) the objective is to classify each cell in the image, where there is a weak image-level classification, while the previous iteration was at the image level

\n

Link to the article that was published on Nature (post the competition): Analysis of Human protein atlas competition

\n

The top ranking models can be found in the model-zoo at BioImage.io

\n

First Place solution by the legendary GM bestfitting

\n

3rd place writeup

\n

4th place writeup by Deiter

\n

5th place writeup

\n

7th place writup

\n

8th place writeup

\n

Realized that this note has been done earlier
\nhttps://www.kaggle.com/c/hpa-single-cell-image-classification/discussion/214527

", "messages": [ { "id": 1173793, "postDate": "2021-01-28T05:51:42.153Z", "content": "

The HPA-1 was held between Dec/Nov 2018-Jan2019. The objective was to classify \"mixed protein cells\" Historically, classification of proteins has been limited to single patterns in one or a few cell types, but in order to fully understand the complexity of the human cell, models must classify mixed patterns across a range of different human cells.

\n

The main difference between the 1st & 2nd HPA competition, is that in the current version (2nd) the objective is to classify each cell in the image, where there is a weak image-level classification, while the previous iteration was at the image level

\n

Link to the article that was published on Nature (post the competition): Analysis of Human protein atlas competition

\n

The top ranking models can be found in the model-zoo at BioImage.io

\n

First Place solution by the legendary GM bestfitting

\n

3rd place writeup

\n

4th place writeup by Deiter

\n

5th place writeup

\n

7th place writup

\n

8th place writeup

\n

Realized that this note has been done earlier
\nhttps://www.kaggle.com/c/hpa-single-cell-image-classification/discussion/214527

", "rawMarkdown": "The HPA-1 was held between Dec/Nov 2018-Jan2019. The objective was to classify \"*mixed protein cells*\" Historically, classification of proteins has been limited to single patterns in one or a few cell types, but in order to fully understand the complexity of the human cell, models must classify mixed patterns across a range of different human cells.\n\n**The main difference between the 1st & 2nd HPA competition, is that in the current version (2nd) the objective is to classify each cell in the image, where there is a weak image-level classification, while the previous iteration was at the image level** \n\nLink to the article that was published on Nature (post the competition): [Analysis of Human protein atlas competition](https://www.nature.com/articles/s41592-019-0658-6)\n\nThe top ranking models can be found in the [model-zoo at BioImage.io](https://bioimage.io/#/?partner=hpa)\n\n[First Place solution by the legendary GM bestfitting](https://www.kaggle.com/c/human-protein-atlas-image-classification/discussion/78109)\n\n[3rd place writeup](https://www.kaggle.com/c/human-protein-atlas-image-classification/discussion/77320)\n\n[4th place writeup by Deiter](https://www.kaggle.com/c/human-protein-atlas-image-classification/discussion/77300)\n\n[5th place writeup](https://www.kaggle.com/c/human-protein-atlas-image-classification/discussion/77731)\n\n[7th place writup](https://www.kaggle.com/c/human-protein-atlas-image-classification/discussion/77269)\n\n[8th place writeup](https://www.kaggle.com/c/human-protein-atlas-image-classification/discussion/77251)\n\n*Realized that this note has been done earlier*\nhttps://www.kaggle.com/c/hpa-single-cell-image-classification/discussion/214527", "votes": 3 } ], "comments": [], "raw_markdown_by_id": { "1173793": "The HPA-1 was held between Dec/Nov 2018-Jan2019. The objective was to classify \"*mixed protein cells*\" Historically, classification of proteins has been limited to single patterns in one or a few cell types, but in order to fully understand the complexity of the human cell, models must classify mixed patterns across a range of different human cells.\n\n**The main difference between the 1st & 2nd HPA competition, is that in the current version (2nd) the objective is to classify each cell in the image, where there is a weak image-level classification, while the previous iteration was at the image level** \n\nLink to the article that was published on Nature (post the competition): [Analysis of Human protein atlas competition](https://www.nature.com/articles/s41592-019-0658-6)\n\nThe top ranking models can be found in the [model-zoo at BioImage.io](https://bioimage.io/#/?partner=hpa)\n\n[First Place solution by the legendary GM bestfitting](https://www.kaggle.com/c/human-protein-atlas-image-classification/discussion/78109)\n\n[3rd place writeup](https://www.kaggle.com/c/human-protein-atlas-image-classification/discussion/77320)\n\n[4th place writeup by Deiter](https://www.kaggle.com/c/human-protein-atlas-image-classification/discussion/77300)\n\n[5th place writeup](https://www.kaggle.com/c/human-protein-atlas-image-classification/discussion/77731)\n\n[7th place writup](https://www.kaggle.com/c/human-protein-atlas-image-classification/discussion/77269)\n\n[8th place writeup](https://www.kaggle.com/c/human-protein-atlas-image-classification/discussion/77251)\n\n*Realized that this note has been done earlier*\nhttps://www.kaggle.com/c/hpa-single-cell-image-classification/discussion/214527" } }