With such complexity and lush rewards, it is natural to seek to build a team.
\n\nSuch as perhaps:
\n\n1 or more Video Specialist(s)\n1 or more Models Specialist(s)\n1 Devious Schemer to anticipate the criminal mind that is being defeated here ;}\n1 Pack leader / Project Manager to keep the rest off useless rabbit trails\n1 bright eyed college student to add insight and advance their skills, you know who you are
\n\nIf you are interested in joining or making a team, why don't you list your skills and see who you might be a match with.
\n\nBest,\nMark
", "messages": [ { "id": "692762", "postDate": "12/11/2019 17:50:39", "content": "With such complexity and lush rewards, it is natural to seek to build a team.
\n\nSuch as perhaps:
\n\n1 or more Video Specialist(s)\n1 or more Models Specialist(s)\n1 Devious Schemer to anticipate the criminal mind that is being defeated here ;}\n1 Pack leader / Project Manager to keep the rest off useless rabbit trails\n1 bright eyed college student to add insight and advance their skills, you know who you are
\n\nIf you are interested in joining or making a team, why don't you list your skills and see who you might be a match with.
\n\nBest,\nMark
", "rawMarkdown": "With such complexity and lush rewards, it is natural to seek to build a team.\n\nSuch as perhaps:\n\n1 or more Video Specialist(s)\n1 or more Models Specialist(s)\n1 Devious Schemer to anticipate the criminal mind that is being defeated here ;}\n1 Pack leader / Project Manager to keep the rest off useless rabbit trails\n1 bright eyed college student to add insight and advance their skills, you know who you are\n\nIf you are interested in joining or making a team, why don't you list your skills and see who you might be a match with.\n\nBest,\nMark", "votes": null }, { "id": "692768", "postDate": "12/11/2019 17:54:09", "content": "Mark Eckdahl:\nI have deep Python knowledge (Django software architect), very Devious with Project Management skills and getting better on Models daily. Plus, I try to create fun. I look forward to crushing fake video culprits with my devious counter-scheming!
", "rawMarkdown": "Mark Eckdahl:\nI have deep Python knowledge (Django software architect), very Devious with Project Management skills and getting better on Models daily. Plus, I try to create fun. I look forward to crushing fake video culprits with my devious counter-scheming!", "votes": null }, { "id": "692777", "postDate": "12/11/2019 18:04:53", "content": "I am a Data science master's student, I participated in multiple competitions, I am still learning, I ve worked on a video project before to detect commercial parts on tv broadcasting, I would like to join a team to learn more !
", "rawMarkdown": "I am a Data science master's student, I participated in multiple competitions, I am still learning, I ve worked on a video project before to detect commercial parts on tv broadcasting, I would like to join a team to learn more !", "votes": null }, { "id": "692854", "postDate": "12/11/2019 19:47:21", "content": "Awesome Adam, you sound like an awesome team member, great fit for the competition.
\n\nYou have a nice paper here: (let me know if you want to look for a team together!)
\n\n", "rawMarkdown": "Awesome Adam, you sound like an awesome team member, great fit for the competition.\n\nYou have a nice paper here: (let me know if you want to look for a team together!)\n\nhttps://towardsdatascience.com/create-and-deploy-an-image-classifier-using-fastai-and-render-in-15-mins-947f9de42d21", "votes": null }, { "id": "692984", "postDate": "12/11/2019 23:37:13", "content": "I'm good at models and have substantial neural network experience but it is somewhat dated (I should be able to brush up very quickly). I tend to use more R but can easily cope with Python. I've done a bit of image work (age determination in faces with HOG, etc.) but no video. I'm very devious and have PM experience but would prefer not to. I suspect that we want an ensemble model with temporal neural networks for feature extraction and then standard models from there. The real trick will be figuring out which features we want to try to find (thus the necessity for deviousness).
", "rawMarkdown": "I'm good at models and have substantial neural network experience but it is somewhat dated (I should be able to brush up very quickly). I tend to use more R but can easily cope with Python. I've done a bit of image work (age determination in faces with HOG, etc.) but no video. I'm very devious and have PM experience but would prefer not to. I suspect that we want an ensemble model with temporal neural networks for feature extraction and then standard models from there. The real trick will be figuring out which features we want to try to find (thus the necessity for deviousness).", "votes": null }, { "id": "693365", "postDate": "12/12/2019 09:55:54", "content": "Python, C#, Tensorflow 2.0 (eagermode), GTX1080ti, and a bit of AWS.\nI'm a former videogame dev.\nI've been playing a lot with VAE, \"small/efficient\" CNN, GANs. I'm a big VAE fan.\nideas for this challenge : \n-detecting MPEG re-encoding artefacts.\n-detecting local color-space aberrations (every camera has a built-in \"white balance\" that changes the limits of the color-space used during recording)\n-detecting fine-grain errors : generated images all suffer from \"smoothness\" in contours, animations...\nLove deviousness.
", "rawMarkdown": "Python, C#, Tensorflow 2.0 (eagermode), GTX1080ti, and a bit of AWS.\nI'm a former videogame dev.\nI've been playing a lot with VAE, \"small/efficient\" CNN, GANs. I'm a big VAE fan.\nideas for this challenge : \n-detecting MPEG re-encoding artefacts.\n-detecting local color-space aberrations (every camera has a built-in \"white balance\" that changes the limits of the color-space used during recording)\n-detecting fine-grain errors : generated images all suffer from \"smoothness\" in contours, animations...\nLove deviousness.", "votes": null }, { "id": "693911", "postDate": "12/12/2019 23:46:22", "content": "Simon, I like where you mind goes, I think more subtle things like encoding are where the best places to discover new detection vulnerabilities.
", "rawMarkdown": "Simon, I like where you mind goes, I think more subtle things like encoding are where the best places to discover new detection vulnerabilities.", "votes": null }, { "id": "693931", "postDate": "12/13/2019 00:58:11", "content": "I suppose focus is important, i.e. I don't want a team who don't have time to do the work, regardless of how skilled they are. I'm digging into the work, and have flexibility in my schedule for committing significant time to it.
", "rawMarkdown": "I suppose focus is important, i.e. I don't want a team who don't have time to do the work, regardless of how skilled they are. I'm digging into the work, and have flexibility in my schedule for committing significant time to it.", "votes": null }, { "id": "713054", "postDate": "01/07/2020 20:55:38", "content": "simon, what if the original video and the fake photo/video has been taken using the same camera or at least camera model. i am just wandering as i am totally newbie and don't know
", "rawMarkdown": "simon, what if the original video and the fake photo/video has been taken using the same camera or at least camera model. i am just wandering as i am totally newbie and don't know", "votes": null } ], "comments": [ { "id": 692768, "author_name": "meckdahl", "author_url": "", "post_date": "12/11/2019 17:54:09", "content": "Mark Eckdahl:\nI have deep Python knowledge (Django software architect), very Devious with Project Management skills and getting better on Models daily. Plus, I try to create fun. I look forward to crushing fake video culprits with my devious counter-scheming!
", "votes": null, "replies": [ { "id": 693931, "author_name": "meckdahl", "author_url": "", "post_date": "12/13/2019 00:58:11", "content": "I suppose focus is important, i.e. I don't want a team who don't have time to do the work, regardless of how skilled they are. I'm digging into the work, and have flexibility in my schedule for committing significant time to it.
", "votes": null, "replies": [] } ] }, { "id": 692777, "author_name": "adamlouly", "author_url": "", "post_date": "12/11/2019 18:04:53", "content": "I am a Data science master's student, I participated in multiple competitions, I am still learning, I ve worked on a video project before to detect commercial parts on tv broadcasting, I would like to join a team to learn more !
", "votes": null, "replies": [ { "id": 692854, "author_name": "meckdahl", "author_url": "", "post_date": "12/11/2019 19:47:21", "content": "Awesome Adam, you sound like an awesome team member, great fit for the competition.
\n\nYou have a nice paper here: (let me know if you want to look for a team together!)
\n\n", "votes": null, "replies": [] } ] }, { "id": 692984, "author_name": "markwaser", "author_url": "", "post_date": "12/11/2019 23:37:13", "content": "I'm good at models and have substantial neural network experience but it is somewhat dated (I should be able to brush up very quickly). I tend to use more R but can easily cope with Python. I've done a bit of image work (age determination in faces with HOG, etc.) but no video. I'm very devious and have PM experience but would prefer not to. I suspect that we want an ensemble model with temporal neural networks for feature extraction and then standard models from there. The real trick will be figuring out which features we want to try to find (thus the necessity for deviousness).
", "votes": null, "replies": [] }, { "id": 693365, "author_name": "simoninparis", "author_url": "", "post_date": "12/12/2019 09:55:54", "content": "Python, C#, Tensorflow 2.0 (eagermode), GTX1080ti, and a bit of AWS.\nI'm a former videogame dev.\nI've been playing a lot with VAE, \"small/efficient\" CNN, GANs. I'm a big VAE fan.\nideas for this challenge : \n-detecting MPEG re-encoding artefacts.\n-detecting local color-space aberrations (every camera has a built-in \"white balance\" that changes the limits of the color-space used during recording)\n-detecting fine-grain errors : generated images all suffer from \"smoothness\" in contours, animations...\nLove deviousness.
", "votes": null, "replies": [ { "id": 693911, "author_name": "meckdahl", "author_url": "", "post_date": "12/12/2019 23:46:22", "content": "Simon, I like where you mind goes, I think more subtle things like encoding are where the best places to discover new detection vulnerabilities.
", "votes": null, "replies": [] }, { "id": 713054, "author_name": "magdyabdulrahman", "author_url": "", "post_date": "01/07/2020 20:55:38", "content": "simon, what if the original video and the fake photo/video has been taken using the same camera or at least camera model. i am just wandering as i am totally newbie and don't know
", "votes": null, "replies": [] } ] } ], "raw_markdown_by_id": { "692762": "With such complexity and lush rewards, it is natural to seek to build a team.\n\nSuch as perhaps:\n\n1 or more Video Specialist(s)\n1 or more Models Specialist(s)\n1 Devious Schemer to anticipate the criminal mind that is being defeated here ;}\n1 Pack leader / Project Manager to keep the rest off useless rabbit trails\n1 bright eyed college student to add insight and advance their skills, you know who you are\n\nIf you are interested in joining or making a team, why don't you list your skills and see who you might be a match with.\n\nBest,\nMark", "692768": "Mark Eckdahl:\nI have deep Python knowledge (Django software architect), very Devious with Project Management skills and getting better on Models daily. Plus, I try to create fun. I look forward to crushing fake video culprits with my devious counter-scheming!", "692777": "I am a Data science master's student, I participated in multiple competitions, I am still learning, I ve worked on a video project before to detect commercial parts on tv broadcasting, I would like to join a team to learn more !", "692854": "Awesome Adam, you sound like an awesome team member, great fit for the competition.\n\nYou have a nice paper here: (let me know if you want to look for a team together!)\n\nhttps://towardsdatascience.com/create-and-deploy-an-image-classifier-using-fastai-and-render-in-15-mins-947f9de42d21", "692984": "I'm good at models and have substantial neural network experience but it is somewhat dated (I should be able to brush up very quickly). I tend to use more R but can easily cope with Python. I've done a bit of image work (age determination in faces with HOG, etc.) but no video. I'm very devious and have PM experience but would prefer not to. I suspect that we want an ensemble model with temporal neural networks for feature extraction and then standard models from there. The real trick will be figuring out which features we want to try to find (thus the necessity for deviousness).", "693365": "Python, C#, Tensorflow 2.0 (eagermode), GTX1080ti, and a bit of AWS.\nI'm a former videogame dev.\nI've been playing a lot with VAE, \"small/efficient\" CNN, GANs. I'm a big VAE fan.\nideas for this challenge : \n-detecting MPEG re-encoding artefacts.\n-detecting local color-space aberrations (every camera has a built-in \"white balance\" that changes the limits of the color-space used during recording)\n-detecting fine-grain errors : generated images all suffer from \"smoothness\" in contours, animations...\nLove deviousness.", "693911": "Simon, I like where you mind goes, I think more subtle things like encoding are where the best places to discover new detection vulnerabilities.", "693931": "I suppose focus is important, i.e. I don't want a team who don't have time to do the work, regardless of how skilled they are. I'm digging into the work, and have flexibility in my schedule for committing significant time to it.", "713054": "simon, what if the original video and the fake photo/video has been taken using the same camera or at least camera model. i am just wandering as i am totally newbie and don't know" }, "source": "meta" }