Hi everyone,
\n\nWe are happy to announce our third annual camera trap challenge, focused on helping to automate species categorization in challenging passive-monitoring cameras placed in the wild and used by ecologists to study animal populations and behavior. The challenge is part of the Seventh Workshop on Fine-Grained Visual Categorization (FGVC7) at CVPR 2020 in Seattle.
\n\nIn our first challenge, iWildCam2018, we focused on binary empty/animal classification. In the second year of our challenge, iWildCam2019, we focused on generalization to an out-of-sample part of the world with a non-identical set of species. This year in iWildCam2020 we are focusing on multimodal data fusion as a way to improve generalization to new camera locations across the globe by providing matched multispectral remote sensing data for each camera trap location, as well as curated subsets of iNaturalist competition data for the species in our dataset.
\n\nLooking forward to the interesting and creative solutions that you all come up with. Feel free to ask questions here, or on the github page!
", "messages": [ { "id": 768287, "postDate": "2020-03-10T15:53:29.930Z", "content": "Hi everyone,
\n\nWe are happy to announce our third annual camera trap challenge, focused on helping to automate species categorization in challenging passive-monitoring cameras placed in the wild and used by ecologists to study animal populations and behavior. The challenge is part of the Seventh Workshop on Fine-Grained Visual Categorization (FGVC7) at CVPR 2020 in Seattle.
\n\nIn our first challenge, iWildCam2018, we focused on binary empty/animal classification. In the second year of our challenge, iWildCam2019, we focused on generalization to an out-of-sample part of the world with a non-identical set of species. This year in iWildCam2020 we are focusing on multimodal data fusion as a way to improve generalization to new camera locations across the globe by providing matched multispectral remote sensing data for each camera trap location, as well as curated subsets of iNaturalist competition data for the species in our dataset.
\n\nLooking forward to the interesting and creative solutions that you all come up with. Feel free to ask questions here, or on the github page!
", "rawMarkdown": "Hi everyone,\n\nWe are happy to announce our third annual camera trap challenge, focused on helping to automate species categorization in challenging passive-monitoring cameras placed in the wild and used by ecologists to study animal populations and behavior. The challenge is part of the [Seventh Workshop on Fine-Grained Visual Categorization](https://sites.google.com/corp/view/fgvc7) (FGVC7) at [CVPR 2020](http://cvpr2020.thecvf.com/) in Seattle.\n\nIn our first challenge, [iWildCam2018](https://www.kaggle.com/c/iwildcam2018), we focused on binary empty/animal classification. In the second year of our challenge, [iWildCam2019](https://www.kaggle.com/c/iwildcam-2019-fgvc6), we focused on generalization to an out-of-sample part of the world with a non-identical set of species. This year in **iWildCam2020** we are focusing on multimodal data fusion as a way to improve generalization to new camera locations across the globe by providing matched multispectral remote sensing data for each camera trap location, as well as curated subsets of iNaturalist competition data for the species in our dataset.\n\nLooking forward to the interesting and creative solutions that you all come up with. Feel free to ask questions here, or on the [github page](https://github.com/visipedia/iwildcam_comp)!", "votes": 11 }, { "id": 1032547, "postDate": "2020-09-30T09:27:28.293Z", "content": "Hi, just here to bump the subject of leaderboard finalization. :)
", "rawMarkdown": "Hi, just here to bump the subject of leaderboard finalization. :)", "votes": 1, "replies": [ { "id": 1032914, "postDate": "2020-09-30T14:08:18.530Z", "content": "Working on it! I've reached out to Kaggle and I'm waiting to hear back.
", "rawMarkdown": "Working on it! I've reached out to Kaggle and I'm waiting to hear back." } ] }, { "id": 1026764, "postDate": "2020-09-25T14:56:12.377Z", "content": "Hi,
\nwhen do you plan to finalize the leaderboard?
Oh dear, I thought I had done this pre-CVPR when we handed out the awards! I'll get right on it :)
", "rawMarkdown": "Oh dear, I thought I had done this pre-CVPR when we handed out the awards! I'll get right on it :)" } ] }, { "id": 771434, "postDate": "2020-03-14T07:03:03.690Z", "content": "Hi \nIn github page, you write\n>We are also providing a general animal detection model which competitors are free to use as they see fit.\n>Sample code for running the detector over a folder of images can be found here.
\n\nIs this mean that we can freely use the detection model for modify and publish in public kernel?\nI want to use part of the code for my kernels.\nThis is first time for me to use such OSS in competition, so please teach me!π
", "rawMarkdown": "Hi \nIn github page, you write\n>We are also providing a general animal detection model which competitors are free to use as they see fit.\n>Sample code for running the detector over a folder of images can be found [here.](https://github.com/Microsoft/CameraTraps/blob/master/detection/run_tf_detector.py)\n\nIs this mean that we can freely use the detection model for modify and publish in public kernel?\nI want to use part of the code for my kernels.\nThis is first time for me to use such OSS in competition, so please teach me!π ", "votes": 1, "replies": [ { "id": 771834, "postDate": "2020-03-14T17:24:49.533Z", "content": "Yep! You can use that detection model in any way you wish. Use it as-is, fine-tune it, whatever you can think of :)
", "rawMarkdown": "Yep! You can use that detection model in any way you wish. Use it as-is, fine-tune it, whatever you can think of :)", "votes": 1, "replies": [ { "id": 2161839, "postDate": "2023-02-27T19:26:40.063Z", "content": "@sbeery link to that model is broken! 404 error.
", "rawMarkdown": "@sbeery link to that model is broken! 404 error." }, { "id": 2161918, "postDate": "2023-02-27T20:29:37.340Z", "content": "That script was renamed and now lives here.
", "rawMarkdown": "That script was renamed and now lives [here](https://github.com/microsoft/CameraTraps/blob/main/detection/run_detector.py)." }, { "id": 2161999, "postDate": "2023-02-27T22:44:25.617Z", "content": "Thank you!
", "rawMarkdown": "Thank you!" } ] }, { "id": 771835, "postDate": "2020-03-14T17:27:11.680Z", "content": "We also already ran the detector over all the images and provided the results, which is a very easy place to stay from since you don't need to run it yourself!
", "rawMarkdown": "We also already ran the detector over all the images and provided the results, which is a very easy place to stay from since you don't need to run it yourself!" }, { "id": 772054, "postDate": "2020-03-15T00:47:30.480Z", "content": "Codemodification is also OK?
", "rawMarkdown": "Codemodification is also OK?" }, { "id": 773829, "postDate": "2020-03-16T00:54:40.483Z", "content": "What code are you referring to?
", "rawMarkdown": "What code are you referring to?", "votes": 1 }, { "id": 775319, "postDate": "2020-03-16T14:54:30.130Z", "content": "These two repositories.\nhttps://github.com/microsoft/CameraTraps\nhttps://github.com/Microsoft/ai4eutils
\n\nBut I found that they are MIT license. Thank you!π
", "rawMarkdown": "These two repositories.\nhttps://github.com/microsoft/CameraTraps\nhttps://github.com/Microsoft/ai4eutils\n\nBut I found that they are MIT license. Thank you!π " }, { "id": 775326, "postDate": "2020-03-16T15:14:33.713Z", "content": "Yeah, use/modify any of that code you'd like :)
", "rawMarkdown": "Yeah, use/modify any of that code you'd like :)" } ] }, { "id": 768723, "postDate": "2020-03-11T05:49:54.127Z", "content": "Hello everyone, I am new to kaggle competition, I am a little confused on how do I start?
", "rawMarkdown": "Hello everyone, I am new to kaggle competition, I am a little confused on how do I start?", "votes": 1, "replies": [ { "id": 769089, "postDate": "2020-03-11T14:20:21.613Z", "content": "The train and test camera trap data are loaded here on kaggle, probably the simplest way to get started is to train a baseline model in a kernel!
", "rawMarkdown": "The train and test camera trap data are loaded here on kaggle, probably the simplest way to get started is to train a baseline model in a kernel!" }, { "id": 769096, "postDate": "2020-03-11T14:28:27.717Z", "content": "Okay are you willing to form a group for the competition?
", "rawMarkdown": "Okay are you willing to form a group for the competition?" }, { "id": 769099, "postDate": "2020-03-11T14:30:52.880Z", "content": "I'm the host! I can't compete :)
", "rawMarkdown": "I'm the host! I can't compete :)" }, { "id": 769340, "postDate": "2020-03-11T19:46:20.147Z", "content": "Hi\nThere is a good YouTube playlist here:\nhttps://www.youtube.com/watch?v=GJBOMWpLpTQ&list=PLqFaTIg4myu8gbDh6oBl7XRYNBlthpDEW
", "rawMarkdown": "Hi\nThere is a good YouTube playlist here:\nhttps://www.youtube.com/watch?v=GJBOMWpLpTQ&list=PLqFaTIg4myu8gbDh6oBl7XRYNBlthpDEW", "votes": 1 } ] }, { "id": 795931, "postDate": "2020-04-03T06:20:35.657Z", "content": "This is a great dataset. I started pre-processing the files over two week ago. I used cv2 and cvlib to find the objects (animals in this case).
\n\nimport cv2\nimport cvlib as cv\nfrom cvlib.object_detection import draw_bbox
\n\nI used the bounding boxes for the objects to create new images. And then used them for training/testing my fast ai model. I seem to be stuck at a score of 0.7. I wish to find some insights. I will be posting my notebooks very soon.
", "rawMarkdown": "This is a great dataset. I started pre-processing the files over two week ago. I used cv2 and cvlib to find the objects (animals in this case). \n\nimport cv2\nimport cvlib as cv\nfrom cvlib.object_detection import draw_bbox\n\nI used the bounding boxes for the objects to create new images. And then used them for training/testing my fast ai model. I seem to be stuck at a score of 0.7. I wish to find some insights. I will be posting my notebooks very soon.", "replies": [ { "id": 804056, "postDate": "2020-04-11T06:58:13.463Z", "content": "Hey I've been following the fastai course too! Could you explain a bit on how did you manage to preprocess and train the model on such a large dataset?
", "rawMarkdown": "Hey I've been following the fastai course too! Could you explain a bit on how did you manage to preprocess and train the model on such a large dataset? " }, { "id": 968435, "postDate": "2020-08-13T02:35:51.307Z", "content": "I'm sorry I wasn't watching this space. It took a lot of time to preprocess the files. I had to let my PC work for the whole day to finish that task. I started out with kaggle notebook but somehow it kept getting disconnected after some interval. When the task was to just resize images, I used office picture manager (v 2007). I know it sounds crazy but I just wanted to get the job done instead of writing code to do it. It is too late but let me know if you need more details.
", "rawMarkdown": "I'm sorry I wasn't watching this space. It took a lot of time to preprocess the files. I had to let my PC work for the whole day to finish that task. I started out with kaggle notebook but somehow it kept getting disconnected after some interval. When the task was to just resize images, I used office picture manager (v 2007). I know it sounds crazy but I just wanted to get the job done instead of writing code to do it. It is too late but let me know if you need more details." } ] }, { "id": 790742, "postDate": "2020-03-29T20:24:18.090Z", "content": "Hi @sbeery : Can you please let me know if the train dataset contains only images from iNatuarlist 2020 or does it also have the images of iNatuarilist 2018 and 2019 along with LANSAT-8 images?
", "rawMarkdown": "Hi @sbeery : Can you please let me know if the train dataset contains only images from iNatuarlist 2020 or does it also have the images of iNatuarilist 2018 and 2019 along with LANSAT-8 images?", "replies": [ { "id": 790962, "postDate": "2020-03-30T01:47:42.390Z", "content": "Hi there - links to download the different components of the training data are available on the competition's GitHub page:
\n\nhttps://github.com/visipedia/iwildcam_comp/#Data
\n\nThe training data includes the camera trap images, the Landsat-8 images, and data from the iNaturalist 2017 and 2018 competitions. Please let us know if this doesn't address your question!
", "rawMarkdown": "Hi there - links to download the different components of the training data are available on the competition's GitHub page:\n\nhttps://github.com/visipedia/iwildcam_comp/#Data\n\nThe training data includes the camera trap images, the Landsat-8 images, and data from the iNaturalist 2017 and 2018 competitions. Please let us know if this doesn't address your question!" } ] }, { "id": 780068, "postDate": "2020-03-20T00:09:47.977Z", "content": "Hi @sbeery , if the CVPR 2020 is cancelled due to the covid19 spread (I hope it doesn't happen) , are these workshop/talks going to continue? For example remotely as at the ICML. I guess you don't have the answer yet, but please keep us informed about that. Thanks!
", "rawMarkdown": "Hi @sbeery , if the CVPR 2020 is cancelled due to the covid19 spread (I hope it doesn't happen) , are these workshop/talks going to continue? For example remotely as at the ICML. I guess you don't have the answer yet, but please keep us informed about that. Thanks!", "replies": [ { "id": 780071, "postDate": "2020-03-20T00:14:22.527Z", "content": "I promise to keep you all informed as information becomes available!
", "rawMarkdown": "I promise to keep you all informed as information becomes available!" } ] }, { "id": 769338, "postDate": "2020-03-11T19:43:49.983Z", "content": "Hi Dr. Beery
\n\nI have some questions about data. Thanks in advance.\nI have used \"annotation\" key of the JSON file to find the labels (which obviously are category_id keys), and the image file names (seems to be the image_id key).\nNow, I don't understand what is id (or image_id, in case of referring id to file names)?
\n\nIt would be nice if you take a little bit time and answer or refer me to a documentation link. I read the readme file but I got more confised:D
", "rawMarkdown": "Hi Dr. Beery\n\nI have some questions about data. Thanks in advance.\nI have used \"annotation\" key of the JSON file to find the labels (which obviously are category_id keys), and the image file names (seems to be the image_id key).\nNow, I don't understand what is id (or image_id, in case of referring id to file names)?\n\nIt would be nice if you take a little bit time and answer or refer me to a documentation link. I read the readme file but I got more confised:D", "replies": [ { "id": 769345, "postDate": "2020-03-11T19:52:45.517Z", "content": "The \"id\" field in the annotation dictionaries is just an identifier for that particular annotation, the \"image_id\" field can be used to match the annotation to the ids in the image dictionaries. You can read more about the COCO dataset format here, we use that format as a base because it's familiar to a lot of computer vision researchers.
", "rawMarkdown": "The \"id\" field in the annotation dictionaries is just an identifier for that particular annotation, the \"image_id\" field can be used to match the annotation to the ids in the image dictionaries. You can read more about the COCO dataset format [here](http://cocodataset.org/#format-data), we use that format as a base because it's familiar to a lot of computer vision researchers.", "votes": 1 } ] }, { "id": 1041918, "postDate": "2020-10-08T01:10:39.620Z", "rawMarkdown": "", "isDeleted": true } ], "comments": [ { "id": 1032547, "author_name": "Mor Kapronczay", "author_url": "", "post_date": "2020-09-30T09:27:28.293000", "content": "Hi, just here to bump the subject of leaderboard finalization. :)
", "votes": 1, "replies": [ { "id": 1032914, "author_name": "sbeery", "author_url": "", "post_date": "2020-09-30T14:08:18.530000", "content": "Working on it! I've reached out to Kaggle and I'm waiting to hear back.
", "votes": 0, "replies": [] } ] }, { "id": 1026764, "author_name": "Winchester", "author_url": "", "post_date": "2020-09-25T14:56:12.377000", "content": "Hi,
\nwhen do you plan to finalize the leaderboard?
Oh dear, I thought I had done this pre-CVPR when we handed out the awards! I'll get right on it :)
", "votes": 0, "replies": [] } ] }, { "id": 771434, "author_name": "Nayu.T.S.", "author_url": "", "post_date": "2020-03-14T07:03:03.690000", "content": "Hi \nIn github page, you write\n>We are also providing a general animal detection model which competitors are free to use as they see fit.\n>Sample code for running the detector over a folder of images can be found here.
\n\nIs this mean that we can freely use the detection model for modify and publish in public kernel?\nI want to use part of the code for my kernels.\nThis is first time for me to use such OSS in competition, so please teach me!π
", "votes": 1, "replies": [ { "id": 771834, "author_name": "sbeery", "author_url": "", "post_date": "2020-03-14T17:24:49.533000", "content": "Yep! You can use that detection model in any way you wish. Use it as-is, fine-tune it, whatever you can think of :)
", "votes": 1, "replies": [ { "id": 2161839, "author_name": "EmadEzzeldin", "author_url": "", "post_date": "2023-02-27T19:26:40.063000", "content": "@sbeery link to that model is broken! 404 error.
", "votes": 0, "replies": [] }, { "id": 2161918, "author_name": "Θtefan Istrate", "author_url": "", "post_date": "2023-02-27T20:29:37.340000", "content": "That script was renamed and now lives here.
", "votes": 0, "replies": [] }, { "id": 2161999, "author_name": "EmadEzzeldin", "author_url": "", "post_date": "2023-02-27T22:44:25.617000", "content": "Thank you!
", "votes": 0, "replies": [] } ] }, { "id": 771835, "author_name": "sbeery", "author_url": "", "post_date": "2020-03-14T17:27:11.680000", "content": "We also already ran the detector over all the images and provided the results, which is a very easy place to stay from since you don't need to run it yourself!
", "votes": 0, "replies": [] }, { "id": 772054, "author_name": "Nayu.T.S.", "author_url": "", "post_date": "2020-03-15T00:47:30.480000", "content": "Codemodification is also OK?
", "votes": 0, "replies": [] }, { "id": 773829, "author_name": "sbeery", "author_url": "", "post_date": "2020-03-16T00:54:40.483000", "content": "What code are you referring to?
", "votes": 1, "replies": [] }, { "id": 775319, "author_name": "Nayu.T.S.", "author_url": "", "post_date": "2020-03-16T14:54:30.130000", "content": "These two repositories.\nhttps://github.com/microsoft/CameraTraps\nhttps://github.com/Microsoft/ai4eutils
\n\nBut I found that they are MIT license. Thank you!π
", "votes": 0, "replies": [] }, { "id": 775326, "author_name": "sbeery", "author_url": "", "post_date": "2020-03-16T15:14:33.713000", "content": "Yeah, use/modify any of that code you'd like :)
", "votes": 0, "replies": [] } ] }, { "id": 768723, "author_name": "Siddharth Bose", "author_url": "", "post_date": "2020-03-11T05:49:54.127000", "content": "Hello everyone, I am new to kaggle competition, I am a little confused on how do I start?
", "votes": 1, "replies": [ { "id": 769089, "author_name": "sbeery", "author_url": "", "post_date": "2020-03-11T14:20:21.613000", "content": "The train and test camera trap data are loaded here on kaggle, probably the simplest way to get started is to train a baseline model in a kernel!
", "votes": 0, "replies": [] }, { "id": 769096, "author_name": "Siddharth Bose", "author_url": "", "post_date": "2020-03-11T14:28:27.717000", "content": "Okay are you willing to form a group for the competition?
", "votes": 0, "replies": [] }, { "id": 769099, "author_name": "sbeery", "author_url": "", "post_date": "2020-03-11T14:30:52.880000", "content": "I'm the host! I can't compete :)
", "votes": 0, "replies": [] }, { "id": 769340, "author_name": "Arvin Nick", "author_url": "", "post_date": "2020-03-11T19:46:20.147000", "content": "Hi\nThere is a good YouTube playlist here:\nhttps://www.youtube.com/watch?v=GJBOMWpLpTQ&list=PLqFaTIg4myu8gbDh6oBl7XRYNBlthpDEW
", "votes": 1, "replies": [] } ] }, { "id": 795931, "author_name": "Manoj Acharya", "author_url": "", "post_date": "2020-04-03T06:20:35.657000", "content": "This is a great dataset. I started pre-processing the files over two week ago. I used cv2 and cvlib to find the objects (animals in this case).
\n\nimport cv2\nimport cvlib as cv\nfrom cvlib.object_detection import draw_bbox
\n\nI used the bounding boxes for the objects to create new images. And then used them for training/testing my fast ai model. I seem to be stuck at a score of 0.7. I wish to find some insights. I will be posting my notebooks very soon.
", "votes": 0, "replies": [ { "id": 804056, "author_name": "Pratyusha Parashar", "author_url": "", "post_date": "2020-04-11T06:58:13.463000", "content": "Hey I've been following the fastai course too! Could you explain a bit on how did you manage to preprocess and train the model on such a large dataset?
", "votes": 0, "replies": [] }, { "id": 968435, "author_name": "Manoj Acharya", "author_url": "", "post_date": "2020-08-13T02:35:51.307000", "content": "I'm sorry I wasn't watching this space. It took a lot of time to preprocess the files. I had to let my PC work for the whole day to finish that task. I started out with kaggle notebook but somehow it kept getting disconnected after some interval. When the task was to just resize images, I used office picture manager (v 2007). I know it sounds crazy but I just wanted to get the job done instead of writing code to do it. It is too late but let me know if you need more details.
", "votes": 0, "replies": [] } ] }, { "id": 790742, "author_name": "DilipMonson", "author_url": "", "post_date": "2020-03-29T20:24:18.090000", "content": "Hi @sbeery : Can you please let me know if the train dataset contains only images from iNatuarlist 2020 or does it also have the images of iNatuarilist 2018 and 2019 along with LANSAT-8 images?
", "votes": 0, "replies": [ { "id": 790962, "author_name": "Elijah Cole", "author_url": "", "post_date": "2020-03-30T01:47:42.390000", "content": "Hi there - links to download the different components of the training data are available on the competition's GitHub page:
\n\nhttps://github.com/visipedia/iwildcam_comp/#Data
\n\nThe training data includes the camera trap images, the Landsat-8 images, and data from the iNaturalist 2017 and 2018 competitions. Please let us know if this doesn't address your question!
", "votes": 0, "replies": [] } ] }, { "id": 780068, "author_name": "Nanashi", "author_url": "", "post_date": "2020-03-20T00:09:47.977000", "content": "Hi @sbeery , if the CVPR 2020 is cancelled due to the covid19 spread (I hope it doesn't happen) , are these workshop/talks going to continue? For example remotely as at the ICML. I guess you don't have the answer yet, but please keep us informed about that. Thanks!
", "votes": 0, "replies": [ { "id": 780071, "author_name": "sbeery", "author_url": "", "post_date": "2020-03-20T00:14:22.527000", "content": "I promise to keep you all informed as information becomes available!
", "votes": 0, "replies": [] } ] }, { "id": 769338, "author_name": "Arvin Nick", "author_url": "", "post_date": "2020-03-11T19:43:49.983000", "content": "Hi Dr. Beery
\n\nI have some questions about data. Thanks in advance.\nI have used \"annotation\" key of the JSON file to find the labels (which obviously are category_id keys), and the image file names (seems to be the image_id key).\nNow, I don't understand what is id (or image_id, in case of referring id to file names)?
\n\nIt would be nice if you take a little bit time and answer or refer me to a documentation link. I read the readme file but I got more confised:D
", "votes": 0, "replies": [ { "id": 769345, "author_name": "sbeery", "author_url": "", "post_date": "2020-03-11T19:52:45.517000", "content": "The \"id\" field in the annotation dictionaries is just an identifier for that particular annotation, the \"image_id\" field can be used to match the annotation to the ids in the image dictionaries. You can read more about the COCO dataset format here, we use that format as a base because it's familiar to a lot of computer vision researchers.
", "votes": 1, "replies": [] } ] }, { "id": 1041918, "author_name": "", "author_url": "", "post_date": "2020-10-08T01:10:39.620000", "content": "", "votes": 0, "replies": [] } ], "raw_markdown_by_id": { "768287": "Hi everyone,\n\nWe are happy to announce our third annual camera trap challenge, focused on helping to automate species categorization in challenging passive-monitoring cameras placed in the wild and used by ecologists to study animal populations and behavior. The challenge is part of the [Seventh Workshop on Fine-Grained Visual Categorization](https://sites.google.com/corp/view/fgvc7) (FGVC7) at [CVPR 2020](http://cvpr2020.thecvf.com/) in Seattle.\n\nIn our first challenge, [iWildCam2018](https://www.kaggle.com/c/iwildcam2018), we focused on binary empty/animal classification. In the second year of our challenge, [iWildCam2019](https://www.kaggle.com/c/iwildcam-2019-fgvc6), we focused on generalization to an out-of-sample part of the world with a non-identical set of species. This year in **iWildCam2020** we are focusing on multimodal data fusion as a way to improve generalization to new camera locations across the globe by providing matched multispectral remote sensing data for each camera trap location, as well as curated subsets of iNaturalist competition data for the species in our dataset.\n\nLooking forward to the interesting and creative solutions that you all come up with. Feel free to ask questions here, or on the [github page](https://github.com/visipedia/iwildcam_comp)!", "1032547": "Hi, just here to bump the subject of leaderboard finalization. :)", "1026764": "Hi, \nwhen do you plan to finalize the leaderboard? ", "771434": "Hi \nIn github page, you write\n>We are also providing a general animal detection model which competitors are free to use as they see fit.\n>Sample code for running the detector over a folder of images can be found [here.](https://github.com/Microsoft/CameraTraps/blob/master/detection/run_tf_detector.py)\n\nIs this mean that we can freely use the detection model for modify and publish in public kernel?\nI want to use part of the code for my kernels.\nThis is first time for me to use such OSS in competition, so please teach me!π ", "768723": "Hello everyone, I am new to kaggle competition, I am a little confused on how do I start?", "795931": "This is a great dataset. I started pre-processing the files over two week ago. I used cv2 and cvlib to find the objects (animals in this case). \n\nimport cv2\nimport cvlib as cv\nfrom cvlib.object_detection import draw_bbox\n\nI used the bounding boxes for the objects to create new images. And then used them for training/testing my fast ai model. I seem to be stuck at a score of 0.7. I wish to find some insights. I will be posting my notebooks very soon.", "790742": "Hi @sbeery : Can you please let me know if the train dataset contains only images from iNatuarlist 2020 or does it also have the images of iNatuarilist 2018 and 2019 along with LANSAT-8 images?", "780068": "Hi @sbeery , if the CVPR 2020 is cancelled due to the covid19 spread (I hope it doesn't happen) , are these workshop/talks going to continue? For example remotely as at the ICML. I guess you don't have the answer yet, but please keep us informed about that. Thanks!", "769338": "Hi Dr. Beery\n\nI have some questions about data. Thanks in advance.\nI have used \"annotation\" key of the JSON file to find the labels (which obviously are category_id keys), and the image file names (seems to be the image_id key).\nNow, I don't understand what is id (or image_id, in case of referring id to file names)?\n\nIt would be nice if you take a little bit time and answer or refer me to a documentation link. I read the readme file but I got more confised:D", "1041918": "" } }