A brief overview of the data makes me skeptical about a direct application of some king of metric learning to the raw image.
\nI would like to share with a kaggle community paper about automatic dolphin fins segmentation using computer vision and a deep learning approach.
\nAbstract:
\nPhoto-identification is a widely used non-invasive technique in biological studies for\nunderstanding if a specimen has been seen multiple times only relying on specific unique visual\ncharacteristics. This information is essential to infer knowledge about the spatial distribution,\nsite fidelity, abundance or habitat use of a species. Today there is a large demand for algorithms that can help domain experts in the analysis of large image datasets. For this reason, it is straightforward that the problem of identify and crop the relevant portion of an image is not negligible in any photo-identification pipeline. This paper approaches the problem of automatically cropping cetaceans images with a hybrid technique based on domain analysis and deep learning. Domain knowledge is applied for proposing relevant regions with the aim of highlighting the dorsal fins, then a binary classification of fin vs. no-fin is performed by a convolutional neural network. Results obtained on real images demonstrate the feasibility of the proposed approach in the automated process of large datasets of Rissoβs dolphins photos, enabling its use on more complex large scale studies. Moreover, the results of this study suggest to extend this methodology to biological investigations of different species.\n\n\nKeywords: photo-identification; cetaceans; Risso; computer vision; deep learning; CNN
\n
Paper link: Combined Color Semantics and Deep Learning for the Automatic Detection of Dolphin Dorsal Fins
", "messages": [ { "id": "1671893", "postDate": "02/01/2022 19:15:42", "content": "A brief overview of the data makes me skeptical about a direct application of some king of metric learning to the raw image.
\nI would like to share with a kaggle community paper about automatic dolphin fins segmentation using computer vision and a deep learning approach.
\nAbstract:
\nPhoto-identification is a widely used non-invasive technique in biological studies for\nunderstanding if a specimen has been seen multiple times only relying on specific unique visual\ncharacteristics. This information is essential to infer knowledge about the spatial distribution,\nsite fidelity, abundance or habitat use of a species. Today there is a large demand for algorithms that can help domain experts in the analysis of large image datasets. For this reason, it is straightforward that the problem of identify and crop the relevant portion of an image is not negligible in any photo-identification pipeline. This paper approaches the problem of automatically cropping cetaceans images with a hybrid technique based on domain analysis and deep learning. Domain knowledge is applied for proposing relevant regions with the aim of highlighting the dorsal fins, then a binary classification of fin vs. no-fin is performed by a convolutional neural network. Results obtained on real images demonstrate the feasibility of the proposed approach in the automated process of large datasets of Rissoβs dolphins photos, enabling its use on more complex large scale studies. Moreover, the results of this study suggest to extend this methodology to biological investigations of different species.\n\n\nKeywords: photo-identification; cetaceans; Risso; computer vision; deep learning; CNN
\n
Paper link: Combined Color Semantics and Deep Learning for the Automatic Detection of Dolphin Dorsal Fins
", "rawMarkdown": "A brief overview of the data makes me skeptical about a direct application of some king of metric learning to the raw image. \n\nI would like to share with a kaggle community paper about automatic dolphin fins segmentation using computer vision and a deep learning approach.\n\n**Abstract:**\n```\nPhoto-identification is a widely used non-invasive technique in biological studies for\nunderstanding if a specimen has been seen multiple times only relying on specific unique visual\ncharacteristics. This information is essential to infer knowledge about the spatial distribution,\nsite fidelity, abundance or habitat use of a species. Today there is a large demand for algorithms that can help domain experts in the analysis of large image datasets. For this reason, it is straightforward that the problem of identify and crop the relevant portion of an image is not negligible in any photo-identification pipeline. This paper approaches the problem of automatically cropping cetaceans images with a hybrid technique based on domain analysis and deep learning. Domain knowledge is applied for proposing relevant regions with the aim of highlighting the dorsal fins, then a binary classification of fin vs. no-fin is performed by a convolutional neural network. Results obtained on real images demonstrate the feasibility of the proposed approach in the automated process of large datasets of Rissoβs dolphins photos, enabling its use on more complex large scale studies. Moreover, the results of this study suggest to extend this methodology to biological investigations of different species.\n```\n\n> Keywords: photo-identification; cetaceans; Risso; computer vision; deep learning; CNN\n\nPaper link: [Combined Color Semantics and Deep Learning for the Automatic Detection of Dolphin Dorsal Fins](https://www.mdpi.com/2079-9292/9/5/758)", "votes": null }, { "id": "1674099", "postDate": "02/03/2022 08:26:06", "content": "\n\nPaper link: Combined Color Semantics and Deep Learning for the Automatic Detection of Dolphin Dorsal Fins
\n
Thank you for sharing, but the link doesn't work, paste there: [Combined Color Semantics and Deep Learning for the Automatic Detection of Dolphin Dorsal Fins](https://www.mdpi.com/2079-9292/9/5/758)
Thank you for pointing me to that issue. Sorry, this is my bad. I have updated the link, so now it works as expected
", "rawMarkdown": "Thank you for pointing me to that issue. Sorry, this is my bad. I have updated the link, so now it works as expected", "votes": null }, { "id": "1674197", "postDate": "02/03/2022 09:47:10", "content": "It is a bug, not your bad.ππ
", "rawMarkdown": "It is a bug, not your bad.ππ", "votes": null }, { "id": "1674206", "postDate": "02/03/2022 09:59:50", "content": "Very Interesting
", "rawMarkdown": "Very Interesting", "votes": null }, { "id": "1674233", "postDate": "02/03/2022 10:37:42", "content": "Definitely, this was my bad. I didn't notice that the paper was downloaded and opened in the browser as a local file from my PC. So, I just copied a link that starts from file:/// π
\n\nPaper link: Combined Color Semantics and Deep Learning for the Automatic Detection of Dolphin Dorsal Fins
\n
Thank you for sharing, but the link doesn't work, paste there: [Combined Color Semantics and Deep Learning for the Automatic Detection of Dolphin Dorsal Fins](https://www.mdpi.com/2079-9292/9/5/758)
Thank you for pointing me to that issue. Sorry, this is my bad. I have updated the link, so now it works as expected
", "votes": null, "replies": [] }, { "id": 1674197, "author_name": "vad13irt", "author_url": "", "post_date": "02/03/2022 09:47:10", "content": "It is a bug, not your bad.ππ
", "votes": null, "replies": [] }, { "id": 1674233, "author_name": "meowmeowmeowmeowmeow", "author_url": "", "post_date": "02/03/2022 10:37:42", "content": "Definitely, this was my bad. I didn't notice that the paper was downloaded and opened in the browser as a local file from my PC. So, I just copied a link that starts from file:/// π
Very Interesting
", "votes": null, "replies": [] } ], "raw_markdown_by_id": { "1671893": "A brief overview of the data makes me skeptical about a direct application of some king of metric learning to the raw image. \n\nI would like to share with a kaggle community paper about automatic dolphin fins segmentation using computer vision and a deep learning approach.\n\n**Abstract:**\n```\nPhoto-identification is a widely used non-invasive technique in biological studies for\nunderstanding if a specimen has been seen multiple times only relying on specific unique visual\ncharacteristics. This information is essential to infer knowledge about the spatial distribution,\nsite fidelity, abundance or habitat use of a species. Today there is a large demand for algorithms that can help domain experts in the analysis of large image datasets. For this reason, it is straightforward that the problem of identify and crop the relevant portion of an image is not negligible in any photo-identification pipeline. This paper approaches the problem of automatically cropping cetaceans images with a hybrid technique based on domain analysis and deep learning. Domain knowledge is applied for proposing relevant regions with the aim of highlighting the dorsal fins, then a binary classification of fin vs. no-fin is performed by a convolutional neural network. Results obtained on real images demonstrate the feasibility of the proposed approach in the automated process of large datasets of Rissoβs dolphins photos, enabling its use on more complex large scale studies. Moreover, the results of this study suggest to extend this methodology to biological investigations of different species.\n```\n\n> Keywords: photo-identification; cetaceans; Risso; computer vision; deep learning; CNN\n\nPaper link: [Combined Color Semantics and Deep Learning for the Automatic Detection of Dolphin Dorsal Fins](https://www.mdpi.com/2079-9292/9/5/758)", "1674099": "> Paper link: Combined Color Semantics and Deep Learning for the Automatic Detection of Dolphin Dorsal Fins\n\nThank you for sharing, but the link doesn't work, paste there: