![\"\"](\"https://www.googleapis.com/download/storage/v1/b/kaggle-forum-message-attachments/o/inbox%2F29434769%2Fcb4d530f7ecfe5b3c675ccaa20431c3e%2FScreenshot_4.png?generation=1773486281991481&alt=media\")
![\"\"](\"https://www.googleapis.com/download/storage/v1/b/kaggle-forum-message-attachments/o/inbox%2F29434769%2Fd3291bffd4df7f75be700dfd135a05ab%2FScreenshot_5.png?generation=1773486288511060&alt=media\")
![\"\"](\"https://www.googleapis.com/download/storage/v1/b/kaggle-forum-message-attachments/o/inbox%2F29434769%2Fe3e242d5544aa02c9ef20b6f7f0169be%2FScreenshot_6.png?generation=1773486295373067&alt=media\")
![\"\"](\"https://www.googleapis.com/download/storage/v1/b/kaggle-forum-message-attachments/o/inbox%2F29434769%2F4a2003893e65ac6a69026b82644f8771%2FScreenshot_7.png?generation=1773486301256963&alt=media\")
Scored below random on this one! Entered with only 21 hours left so I never got past color into actual machine learning. But I'm proud of the data work so I wanted to share.
\nI built a color correction pipeline that histogram-matches every image per identity to a reference photo, then used k-means (k=2) to separate fur and spot layers independently. Each jaguar gets its own canonical fur RGB and spot RGB signature. Cleaned out SAM3 green/teal segmentation artifacts before any color extraction. Never made it to shape analysis, just color.
\nThe separated layers look cool though so here they are. I'll be continuing to work with this dataset over the next couple weeks as my capstone for my master's program. Good luck everyone!
", "messages": [ { "id": 3420980, "postDate": "2026-03-14T11:08:08.880Z", "content": "
Scored below random on this one! Entered with only 21 hours left so I never got past color into actual machine learning. But I'm proud of the data work so I wanted to share.
\nI built a color correction pipeline that histogram-matches every image per identity to a reference photo, then used k-means (k=2) to separate fur and spot layers independently. Each jaguar gets its own canonical fur RGB and spot RGB signature. Cleaned out SAM3 green/teal segmentation artifacts before any color extraction. Never made it to shape analysis, just color.
\nThe separated layers look cool though so here they are. I'll be continuing to work with this dataset over the next couple weeks as my capstone for my master's program. Good luck everyone!
", "rawMarkdown": "\n\n\n\n\n\n\n\n\n\n\nScored below random on this one! Entered with only 21 hours left so I never got past color into actual machine learning. But I'm proud of the data work so I wanted to share. \n\nI built a color correction pipeline that histogram-matches every image per identity to a reference photo, then used k-means (k=2) to separate fur and spot layers independently. Each jaguar gets its own canonical fur RGB and spot RGB signature. Cleaned out SAM3 green/teal segmentation artifacts before any color extraction. Never made it to shape analysis, just color.\n\n The separated layers look cool though so here they are. I'll be continuing to work with this dataset over the next couple weeks as my capstone for my master's program. Good luck everyone! \n", "votes": 1 } ], "comments": [], "raw_markdown_by_id": { "3420980": "\n\n\n\n\n\n\n\n\n\n\nScored below random on this one! Entered with only 21 hours left so I never got past color into actual machine learning. But I'm proud of the data work so I wanted to share. \n\nI built a color correction pipeline that histogram-matches every image per identity to a reference photo, then used k-means (k=2) to separate fur and spot layers independently. Each jaguar gets its own canonical fur RGB and spot RGB signature. Cleaned out SAM3 green/teal segmentation artifacts before any color extraction. Never made it to shape analysis, just color.\n\n The separated layers look cool though so here they are. I'll be continuing to work with this dataset over the next couple weeks as my capstone for my master's program. Good luck everyone! \n" } }