I treat this problem as a segmentation task with MSE loss. My base model is SwinUnetr from monai framework. A few data augmentation technics are used: random flip, gaussian noise, grid distortion.
\nAn input is resized to 224, but the loss is calculated on original size image after resizing. The solutions compose of 2 stages.
\nThe first one uses wind speed (ustar), temperature (theta), and fire-line location (xi), slope (alpha) and mean wind speed (u10) of first 5 time steps to predict 20 next line location (17 input channels, 20 output channels).
\nThe second stage uses previous 25 fire-line location, alpha, u10 to predict the next 25 fire-line location (27 input channels, 25 output channels).
\nAn output of the previous prediction is in range 0-1. In order to binarize and enhance the signal, otsu threshold per frame is applied. I apply post-processing with use of otsu threshold and a binary dilation to zero background (places where is no fire). The final output are floats in range 0 - 1.
There are still many options to check:
\nI hope this post makes me discussion expert :)
", "comments": [ { "id": 3027434, "author_name": "Zhuoqun Li", "author_url": "", "post_date": "2024-10-24T19:58:43.100000", "content": "Great job! Thank you.
", "votes": 1, "replies": [] } ], "comment_count": 1, "messages": [ { "id": 3022030, "postDate": "2024-10-19T07:19:53.737Z", "content": "I treat this problem as a segmentation task with MSE loss. My base model is SwinUnetr from monai framework. A few data augmentation technics are used: random flip, gaussian noise, grid distortion.
\nAn input is resized to 224, but the loss is calculated on original size image after resizing. The solutions compose of 2 stages.
\nThe first one uses wind speed (ustar), temperature (theta), and fire-line location (xi), slope (alpha) and mean wind speed (u10) of first 5 time steps to predict 20 next line location (17 input channels, 20 output channels).
\nThe second stage uses previous 25 fire-line location, alpha, u10 to predict the next 25 fire-line location (27 input channels, 25 output channels).
\nAn output of the previous prediction is in range 0-1. In order to binarize and enhance the signal, otsu threshold per frame is applied. I apply post-processing with use of otsu threshold and a binary dilation to zero background (places where is no fire). The final output are floats in range 0 - 1.
There are still many options to check:
\nI hope this post makes me discussion expert :)
", "rawMarkdown": "I treat this problem as a segmentation task with MSE loss. My base model is SwinUnetr from monai framework. A few data augmentation technics are used: random flip, gaussian noise, grid distortion.\nAn input is resized to 224, but the loss is calculated on original size image after resizing. The solutions compose of 2 stages.\nThe first one uses wind speed (ustar), temperature (theta), and fire-line location (xi), slope (alpha) and mean wind speed (u10) of first 5 time steps to predict 20 next line location (17 input channels, 20 output channels).\nThe second stage uses previous 25 fire-line location, alpha, u10 to predict the next 25 fire-line location (27 input channels, 25 output channels).\nAn output of the previous prediction is in range 0-1. In order to binarize and enhance the signal, otsu threshold per frame is applied. I apply post-processing with use of otsu threshold and a binary dilation to zero background (places where is no fire). The final output are floats in range 0 - 1. \n\n[Notebook](https://www.kaggle.com/code/jankowalski2000/segmentation)\n\nThere are still many options to check:\n1. dice loss, tversky loss, combo loss with MSE\n2. more aggressive augmentation methods\n3. learning rate schedulers\n\nI hope this post makes me discussion expert :)\n", "votes": 5 }, { "id": 3027434, "postDate": "2024-10-24T19:58:43.100Z", "content": "Great job! Thank you.
", "rawMarkdown": "Great job! Thank you.", "votes": 1 } ], "complete": true, "source": "live", "fetched_at": "2026-10-19T13:44:27Z", "competition_id": 85210, "competition_slug": "2024-flame-ai-challenge", "team_id": 12670834, "solution_url": "/competitions/2024-flame-ai-challenge/writeups/rafa-paw-owski-segmentation-approach" }