{ "id": 304525, "title": "Welcome to whale and dolphin photo-identification", "url": "/competitions/happy-whale-and-dolphin/discussion/304525", "author_name": "Ted Cheeseman", "post_date": "2022-02-01T18:11:13.212000", "votes": 58, "comment_count": 0, "views": 0, "content": "

Welcome to the Whale and Dolphin Identification Research Prediction Competition

\n

As competition host, I am extremely excited about this effort, based on the incredibly successful result of our previous competition. In that effort, we implemented the algorithm from the 3rd place winner @pudae81, resulting in a system able to identify any humpback whale anywhere in the world, accessible to all via Happywhale -- and a publication about the algorithm. The system is changing science and very much benefitting marine conservation -- we have identified most living humpback whales in the North Pacific. From this, we are building a better understanding of the health of humpback whale populations. Just this past week (late January 2022) we published a study demonstrating the impact of entanglement in fishing gear along the US West Coast, a very important step in saving whales from horrible death by laceration and drowning by commercial fishing gear.

\n

So, why a new competition? To properly care for our oceans, we need to be able to identify individuals of other whale and dolphin species. The body part most readily visible in most species is the dorsal fin or the back when they surface to breathe. Not all species have dorsal fins and there are a few different ways and different features to use for individual identification. With such a success to build upon, we know so much more is possible. Most whales and dolphins are individually recognizable from photographs, \"photo-ID\", an extremely powerful technique but one that is limited by the time, expertise and memory required to recognize individuals by eye. In a population of 10s or 100s of animals, this is not so difficult but in a dataset of thousands (there are 66,000 humpbacks in our dataset now!) this becomes impossible.

\n

Here we have gathered a large and diverse dataset -- I will create a series of posts displaying some of the features used for ID. Some species are easy with good quality photos, but much harder with poorer quality photos. Other species, like beluga whales with no dorsal fins, are difficult unless the individual has prominent scarring -- we invite you to do your best, knowing that the best outcomes will be put to use to improve conservation science in wonderfully innovative ways.

\n

Good luck!
\nTed Cheeseman, collaborators, and all of us at Happywhale

\n

Example of a photo-ID match for images of a Sei Whale

\n

\"Sei

\n

Source: Weir, C.R. (ed.). (2022). Conserving Falklands' whale populations: addressing data deficiencies for informed management. Technical Report for DPLUS082. Falklands Conservation, Stanley, Falkland Islands. Version 1, 31 Jan 2022. 210 pp.

", "messages": [ { "id": 1671784, "postDate": "2022-02-01T18:11:13.213Z", "content": "

Welcome to the Whale and Dolphin Identification Research Prediction Competition

\n

As competition host, I am extremely excited about this effort, based on the incredibly successful result of our previous competition. In that effort, we implemented the algorithm from the 3rd place winner @pudae81, resulting in a system able to identify any humpback whale anywhere in the world, accessible to all via Happywhale -- and a publication about the algorithm. The system is changing science and very much benefitting marine conservation -- we have identified most living humpback whales in the North Pacific. From this, we are building a better understanding of the health of humpback whale populations. Just this past week (late January 2022) we published a study demonstrating the impact of entanglement in fishing gear along the US West Coast, a very important step in saving whales from horrible death by laceration and drowning by commercial fishing gear.

\n

So, why a new competition? To properly care for our oceans, we need to be able to identify individuals of other whale and dolphin species. The body part most readily visible in most species is the dorsal fin or the back when they surface to breathe. Not all species have dorsal fins and there are a few different ways and different features to use for individual identification. With such a success to build upon, we know so much more is possible. Most whales and dolphins are individually recognizable from photographs, \"photo-ID\", an extremely powerful technique but one that is limited by the time, expertise and memory required to recognize individuals by eye. In a population of 10s or 100s of animals, this is not so difficult but in a dataset of thousands (there are 66,000 humpbacks in our dataset now!) this becomes impossible.

\n

Here we have gathered a large and diverse dataset -- I will create a series of posts displaying some of the features used for ID. Some species are easy with good quality photos, but much harder with poorer quality photos. Other species, like beluga whales with no dorsal fins, are difficult unless the individual has prominent scarring -- we invite you to do your best, knowing that the best outcomes will be put to use to improve conservation science in wonderfully innovative ways.

\n

Good luck!
\nTed Cheeseman, collaborators, and all of us at Happywhale

\n

Example of a photo-ID match for images of a Sei Whale

\n

\"Sei

\n

Source: Weir, C.R. (ed.). (2022). Conserving Falklands' whale populations: addressing data deficiencies for informed management. Technical Report for DPLUS082. Falklands Conservation, Stanley, Falkland Islands. Version 1, 31 Jan 2022. 210 pp.

", "rawMarkdown": "Welcome to the Whale and Dolphin Identification Research Prediction Competition\n\nAs competition host, I am extremely excited about this effort, based on the incredibly successful result of [**our previous competition**](https://www.kaggle.com/c/humpback-whale-identification). In that effort, we implemented the algorithm from the 3rd place winner @pudae81, resulting in a system able to identify any humpback whale anywhere in the world, accessible to all via [**Happywhale**](https://happywhale.com/) -- and a [**publication about the algorithm**](https://rdcu.be/cCOtw). The system is changing science and very much benefitting marine conservation -- we have identified most living humpback whales in the North Pacific. From this, we are building a better understanding of the health of humpback whale populations. Just this past week (late January 2022) we published [**a study demonstrating the impact of entanglement**](https://www.frontiersin.org/articles/10.3389/fmars.2021.779448/full) in fishing gear along the US West Coast, a very important step in saving whales from horrible death by laceration and drowning by commercial fishing gear.\n\nSo, why a new competition? To properly care for our oceans, we **need** to be able to identify individuals of other whale and dolphin species. The body part most readily visible in most species is the dorsal fin or the back when they surface to breathe. Not all species have dorsal fins and there are a few different ways and different features to use for individual identification. With such a success to build upon, we know so much more is possible. Most whales and dolphins are individually recognizable from photographs, \"photo-ID\", an extremely powerful technique but one that is limited by the time, expertise and memory required to recognize individuals by eye. In a population of 10s or 100s of animals, this is not so difficult but in a dataset of thousands (there are 66,000 humpbacks in our dataset now!) this becomes impossible. \n\nHere we have gathered a large and diverse dataset -- I will create a series of posts displaying some of the features used for ID. Some species are easy with good quality photos, but much harder with poorer quality photos. Other species, like beluga whales with no dorsal fins, are difficult unless the individual has prominent scarring -- we invite you to do your best, knowing that the best outcomes will be put to use to improve conservation science in wonderfully innovative ways.\n\nGood luck!\nTed Cheeseman, collaborators, and all of us at Happywhale\n\nExample of a photo-ID match for images of a Sei Whale\n\n\"Sei\n\nSource: Weir, C.R. (ed.). (2022). Conserving Falklands' whale populations: addressing data deficiencies for informed management. Technical Report for DPLUS082. Falklands Conservation, Stanley, Falkland Islands. Version 1, 31 Jan 2022. 210 pp.\n", "votes": 56 } ], "comments": [], "raw_markdown_by_id": { "1671784": "Welcome to the Whale and Dolphin Identification Research Prediction Competition\n\nAs competition host, I am extremely excited about this effort, based on the incredibly successful result of [**our previous competition**](https://www.kaggle.com/c/humpback-whale-identification). In that effort, we implemented the algorithm from the 3rd place winner @pudae81, resulting in a system able to identify any humpback whale anywhere in the world, accessible to all via [**Happywhale**](https://happywhale.com/) -- and a [**publication about the algorithm**](https://rdcu.be/cCOtw). The system is changing science and very much benefitting marine conservation -- we have identified most living humpback whales in the North Pacific. From this, we are building a better understanding of the health of humpback whale populations. Just this past week (late January 2022) we published [**a study demonstrating the impact of entanglement**](https://www.frontiersin.org/articles/10.3389/fmars.2021.779448/full) in fishing gear along the US West Coast, a very important step in saving whales from horrible death by laceration and drowning by commercial fishing gear.\n\nSo, why a new competition? To properly care for our oceans, we **need** to be able to identify individuals of other whale and dolphin species. The body part most readily visible in most species is the dorsal fin or the back when they surface to breathe. Not all species have dorsal fins and there are a few different ways and different features to use for individual identification. With such a success to build upon, we know so much more is possible. Most whales and dolphins are individually recognizable from photographs, \"photo-ID\", an extremely powerful technique but one that is limited by the time, expertise and memory required to recognize individuals by eye. In a population of 10s or 100s of animals, this is not so difficult but in a dataset of thousands (there are 66,000 humpbacks in our dataset now!) this becomes impossible. \n\nHere we have gathered a large and diverse dataset -- I will create a series of posts displaying some of the features used for ID. Some species are easy with good quality photos, but much harder with poorer quality photos. Other species, like beluga whales with no dorsal fins, are difficult unless the individual has prominent scarring -- we invite you to do your best, knowing that the best outcomes will be put to use to improve conservation science in wonderfully innovative ways.\n\nGood luck!\nTed Cheeseman, collaborators, and all of us at Happywhale\n\nExample of a photo-ID match for images of a Sei Whale\n\n\"Sei\n\nSource: Weir, C.R. (ed.). (2022). Conserving Falklands' whale populations: addressing data deficiencies for informed management. Technical Report for DPLUS082. Falklands Conservation, Stanley, Falkland Islands. Version 1, 31 Jan 2022. 210 pp.\n" } }