This is a python script . After separating the color channels into principal components, it uses a combination of 1) finding the pixels with the largest difference from the median of the whole image and 2) gridding the image and comparing the histograms between the histogram of each section and the histogram of the entire image using Earth Mover's Distance.
\n\nThe script takes an image location and returns the coordinates for an estimated line that goes from the whale's head to tail.
\n\nBonus is that this script does not use OpenCV. It does use pyemd, scikit image, scikit learn, numpy, and scipy.
", "messages": [ { "id": "103650", "postDate": "01/05/2016 06:13:34", "content": "This is a python script . After separating the color channels into principal components, it uses a combination of 1) finding the pixels with the largest difference from the median of the whole image and 2) gridding the image and comparing the histograms between the histogram of each section and the histogram of the entire image using Earth Mover's Distance.
\n\nThe script takes an image location and returns the coordinates for an estimated line that goes from the whale's head to tail.
\n\nBonus is that this script does not use OpenCV. It does use pyemd, scikit image, scikit learn, numpy, and scipy.
", "rawMarkdown": "This is a python script . After separating the color channels into principal components, it uses a combination of 1) finding the pixels with the largest difference from the median of the whole image and 2) gridding the image and comparing the histograms between the histogram of each section and the histogram of the entire image using Earth Mover's Distance.\r\n\r\nThe script takes an image location and returns the coordinates for an estimated line that goes from the whale's head to tail.\r\n\r\nBonus is that this script does not use OpenCV. It does use [pyemd][1], scikit image, scikit learn, numpy, and scipy.\r\n\r\n\r\n [1]: https://pypi.python.org/pypi/pyemd/0.0.7 \"pyEMD\"", "votes": null }, { "id": "103676", "postDate": "01/05/2016 13:36:11", "content": "Thanks KA for the code. Works fine. Could you please share the reference is based on?
", "rawMarkdown": "Thanks KA for the code. Works fine. Could you please share the reference is based on?", "votes": null }, { "id": "103682", "postDate": "01/05/2016 14:41:14", "content": "good contribution KA! I tested it with a few images and obtained good results.
", "rawMarkdown": "good contribution KA! I tested it with a few images and obtained good results.", "votes": null }, { "id": "103698", "postDate": "01/05/2016 16:23:34", "content": "No reference. PCA is used to reduce dimensionality from 3 to 2 and remove noise. The median difference seemed logical because usually there is more water than whale. The Earth Mover Distance is just a higher order measure which takes into account the shapes of the histograms.
\n\nThe values of the parameters were arrived at using pretty standard ML.
", "rawMarkdown": "No reference. PCA is used to reduce dimensionality from 3 to 2 and remove noise. The median difference seemed logical because usually there is more water than whale. The Earth Mover Distance is just a higher order measure which takes into account the shapes of the histograms.\r\n\r\nThe values of the parameters were arrived at using pretty standard ML.", "votes": null } ], "comments": [ { "id": 103676, "author_name": "potamitis", "author_url": "", "post_date": "01/05/2016 13:36:11", "content": "Thanks KA for the code. Works fine. Could you please share the reference is based on?
", "votes": null, "replies": [] }, { "id": 103682, "author_name": "debypoli", "author_url": "", "post_date": "01/05/2016 14:41:14", "content": "good contribution KA! I tested it with a few images and obtained good results.
", "votes": null, "replies": [] }, { "id": 103698, "author_name": "", "author_url": "", "post_date": "01/05/2016 16:23:34", "content": "No reference. PCA is used to reduce dimensionality from 3 to 2 and remove noise. The median difference seemed logical because usually there is more water than whale. The Earth Mover Distance is just a higher order measure which takes into account the shapes of the histograms.
\n\nThe values of the parameters were arrived at using pretty standard ML.
", "votes": null, "replies": [] } ], "raw_markdown_by_id": { "103650": "This is a python script . After separating the color channels into principal components, it uses a combination of 1) finding the pixels with the largest difference from the median of the whole image and 2) gridding the image and comparing the histograms between the histogram of each section and the histogram of the entire image using Earth Mover's Distance.\r\n\r\nThe script takes an image location and returns the coordinates for an estimated line that goes from the whale's head to tail.\r\n\r\nBonus is that this script does not use OpenCV. It does use [pyemd][1], scikit image, scikit learn, numpy, and scipy.\r\n\r\n\r\n [1]: https://pypi.python.org/pypi/pyemd/0.0.7 \"pyEMD\"", "103676": "Thanks KA for the code. Works fine. Could you please share the reference is based on?", "103682": "good contribution KA! I tested it with a few images and obtained good results.", "103698": "No reference. PCA is used to reduce dimensionality from 3 to 2 and remove noise. The median difference seemed logical because usually there is more water than whale. The Earth Mover Distance is just a higher order measure which takes into account the shapes of the histograms.\r\n\r\nThe values of the parameters were arrived at using pretty standard ML." }, "source": "meta" }