{ "id": 120076, "title": "The algorithm that Competition Organizer-Baidu chooses !", "url": "/competitions/pku-autonomous-driving/discussion/120076", "author_name": "DiegoJohnson", "post_date": "2019-12-03T13:27:55.867000", "votes": 55, "comment_count": 14, "views": 0, "content": "

Baidu Apollo

\n\n

Baidu Apollo has compared some different algorithms, and finally choose YOLO as main algorithm, modified DarkNet as Feature Extractor, because they have high requirement in speed.

\n\n

\"1\"

\n\n

They call this modified algorithm 'Multi task YOLO-3D'.

\n\n

\"2\"

\n\n

The parameters that be predicted:

\n\n

\"3\"\n\"4\"

\n\n

then they simplify the parameters, but we need to predict 'yaw, pitch, roll, x, y, z', and no D(length, width, height)

\n\n

\"5\"

\n\n

It seems yaw is more important than pitch, roll.

\n\n

\"6\"

\n\n

Train

\n\n

\"7\"\n\"8\"

\n\n

We don't need the distance, but this picture is interesting.

\n\n

\"9\"

\n\n

Whole pipeline:

\n\n

\"10\"

\n\n

It takes me a lot of time, if you think it's useful, please give me a upvote, thanks.

\n\n

These are my other topics.\nThe coordinate system of this competition\nAlgorithm Selection for Beginner!\nUnderstand Camera Intrinsic Parameters and X, Y, Z\nAlgorithm Selection in 6D Pose Estimation\nHow the Competition Organizer get train.csv !\nSource of 6D Pose Estimation\nA way to improve the accuracy of model\nObject Detection in 20 Years

\n\n

My notebook:\nCenternet - Objects as Points\nA Way to Regress Translation and Rotation\nBest Algorithm so far - Google's new paper in Nov\nA better way to regress Yaw!\nA clear view of car pose\nDataset with preprocess

", "messages": [ { "id": 686742, "postDate": "2019-12-03T13:27:55.867Z", "content": "

Baidu Apollo

\n\n

Baidu Apollo has compared some different algorithms, and finally choose YOLO as main algorithm, modified DarkNet as Feature Extractor, because they have high requirement in speed.

\n\n

\"1\"

\n\n

They call this modified algorithm 'Multi task YOLO-3D'.

\n\n

\"2\"

\n\n

The parameters that be predicted:

\n\n

\"3\"\n\"4\"

\n\n

then they simplify the parameters, but we need to predict 'yaw, pitch, roll, x, y, z', and no D(length, width, height)

\n\n

\"5\"

\n\n

It seems yaw is more important than pitch, roll.

\n\n

\"6\"

\n\n

Train

\n\n

\"7\"\n\"8\"

\n\n

We don't need the distance, but this picture is interesting.

\n\n

\"9\"

\n\n

Whole pipeline:

\n\n

\"10\"

\n\n

It takes me a lot of time, if you think it's useful, please give me a upvote, thanks.

\n\n

These are my other topics.\nThe coordinate system of this competition\nAlgorithm Selection for Beginner!\nUnderstand Camera Intrinsic Parameters and X, Y, Z\nAlgorithm Selection in 6D Pose Estimation\nHow the Competition Organizer get train.csv !\nSource of 6D Pose Estimation\nA way to improve the accuracy of model\nObject Detection in 20 Years

\n\n

My notebook:\nCenternet - Objects as Points\nA Way to Regress Translation and Rotation\nBest Algorithm so far - Google's new paper in Nov\nA better way to regress Yaw!\nA clear view of car pose\nDataset with preprocess

", "rawMarkdown": "## Baidu Apollo\n\nBaidu Apollo has compared some different algorithms, and finally choose **YOLO** as main algorithm, modified **DarkNet** as Feature Extractor, because they have high requirement in speed.\n\n![1](http://file.elecfans.com/web1/M00/66/89/o4YBAFu9o3OAUdNsAAC81sWxRlE367.jpg)\n\n## They call this modified algorithm 'Multi task YOLO-3D'.\n\n![2](http://file.elecfans.com/web1/M00/66/89/o4YBAFu9o3SAHlJtAADazbyMU68795.jpg)\n\n## The parameters that be predicted:\n\n![3](http://file.elecfans.com/web1/M00/66/89/o4YBAFu9o3eAClnHAADNMJxXt00919.jpg)\n![4](http://file.elecfans.com/web1/M00/66/89/o4YBAFu9o3iAR2b4AADGD6L6W68535.jpg)\n\n## then they simplify the parameters, but we need to predict 'yaw, pitch, roll, x, y, z', and no D(length, width, height)\n\n![5](http://file.elecfans.com/web1/M00/66/89/o4YBAFu9o3mAPSYZAAD0_fFm_tw775.jpg)\n\n## It seems yaw is more important than pitch, roll.\n\n![6](http://file.elecfans.com/web1/M00/66/89/o4YBAFu9o3qAVgU5AADwEa89d3s458.jpg)\n\n## Train\n\n![7](http://file.elecfans.com/web1/M00/66/89/o4YBAFu9o3qAInKVAACk-xcwd70828.jpg)\n![8](http://file.elecfans.com/web1/M00/66/89/o4YBAFu9o3yAPT-HAAD5UysFgO0007.jpg)\n\n## We don't need the distance, but this picture is interesting.\n![9](http://file.elecfans.com/web1/M00/66/89/o4YBAFu9o36AJL0rAADst_qutg8528.jpg)\n\n## Whole pipeline:\n![10](http://file.elecfans.com/web1/M00/66/89/o4YBAFu9o4GAfNBjAACgfNjgGt8637.jpg)\n\n## It takes me a lot of time, if you think it's useful, please give me a upvote, thanks.\n\nThese are my other topics.\n[The coordinate system of this competition](https://www.kaggle.com/c/pku-autonomous-driving/discussion/123385)\n[Algorithm Selection for Beginner!](https://www.kaggle.com/c/pku-autonomous-driving/discussion/120015#latest-686515)\n[Understand Camera Intrinsic Parameters and X, Y, Z](https://www.kaggle.com/c/pku-autonomous-driving/discussion/120083#latest-687359)\n[Algorithm Selection in 6D Pose Estimation](https://www.kaggle.com/c/pku-autonomous-driving/discussion/120381)\n[How the Competition Organizer get train.csv !](https://www.kaggle.com/c/pku-autonomous-driving/discussion/120443)\n[Source of 6D Pose Estimation](https://www.kaggle.com/c/pku-autonomous-driving/discussion/120582)\n[A way to improve the accuracy of model](https://www.kaggle.com/c/pku-autonomous-driving/discussion/120653)\n[Object Detection in 20 Years](https://www.kaggle.com/c/pku-autonomous-driving/discussion/120710)\n\n\nMy notebook:\n[Centernet - Objects as Points](https://www.kaggle.com/diegojohnson/centernet-objects-as-points)\n[A Way to Regress Translation and Rotation](https://www.kaggle.com/diegojohnson/a-way-to-regress-translation-and-rotation?scriptVersionId=24677131)\n[Best Algorithm so far - Google's new paper in Nov](https://www.kaggle.com/diegojohnson/best-algorithm-so-far-google-s-new-paper-in-nov)\n[A better way to regress Yaw!](https://www.kaggle.com/diegojohnson/a-better-way-to-regress-yaw)\n[A clear view of car pose](https://www.kaggle.com/diegojohnson/a-clear-view-of-car-pose)\n[Dataset with preprocess](https://www.kaggle.com/c/pku-autonomous-driving/discussion/124480)", "votes": 55 }, { "id": 686745, "postDate": "2019-12-03T13:32:20.570Z", "content": "

Oh man .. you are on a roll ! Pitch ,yaw too ..

", "rawMarkdown": "Oh man .. you are on a roll ! Pitch ,yaw too ..", "votes": 3, "replies": [ { "id": 686749, "postDate": "2019-12-03T13:37:59.833Z", "content": "

Hope you like it.

", "rawMarkdown": "Hope you like it.", "votes": 1 }, { "id": 686759, "postDate": "2019-12-03T13:55:32.360Z", "content": "

ofcourse , great find and great effort .

", "rawMarkdown": "ofcourse , great find and great effort .", "votes": 1 } ] }, { "id": 689285, "postDate": "2019-12-06T18:49:08.107Z", "content": "

Very informative and helpful.

", "rawMarkdown": "Very informative and helpful.", "votes": 1, "replies": [ { "id": 689511, "postDate": "2019-12-07T01:52:19.683Z", "content": "

thanks. I will keep updata, as I find useful information, you also can see my other topics.

", "rawMarkdown": "thanks. I will keep updata, as I find useful information, you also can see my other topics.", "votes": 1 } ] }, { "id": 688993, "postDate": "2019-12-06T10:19:18.613Z", "content": "

reference for me.

", "rawMarkdown": "reference for me.", "votes": 1 }, { "id": 687346, "postDate": "2019-12-04T09:20:09.270Z", "content": "

Great Find\nVery Helpful\nThanks @diegojohnson

", "rawMarkdown": "Great Find\nVery Helpful\nThanks @diegojohnson ", "votes": 1, "replies": [ { "id": 687362, "postDate": "2019-12-04T09:39:27.043Z", "content": "

Thanks😄

", "rawMarkdown": "Thanks😄 " } ] }, { "id": 687282, "postDate": "2019-12-04T07:10:18.170Z", "content": "

Thank for your sharing. Can you kindly give us the link to this ppt?

", "rawMarkdown": "Thank for your sharing. Can you kindly give us the link to this ppt?", "votes": 1, "replies": [ { "id": 687285, "postDate": "2019-12-04T07:17:51.080Z", "content": "

It's chinese, and I don't have ppt file.\nhttp://www.sohu.com/a/258872057_613224

", "rawMarkdown": "It's chinese, and I don't have ppt file.\nhttp://www.sohu.com/a/258872057_613224", "votes": 1 } ] }, { "id": 686758, "postDate": "2019-12-03T13:53:16.817Z", "content": "

Great find

", "rawMarkdown": "Great find ", "votes": 2, "replies": [ { "id": 686763, "postDate": "2019-12-03T14:00:47.633Z", "content": "

The source is chinese, so it's hard to find.😂

", "rawMarkdown": "The source is chinese, so it's hard to find.😂 " } ] }, { "id": 692335, "postDate": "2019-12-11T06:40:38.060Z", "content": "

Striking Post. You got a new follower ..

", "rawMarkdown": "Striking Post. You got a new follower ..", "replies": [ { "id": 692403, "postDate": "2019-12-11T08:15:52.263Z", "rawMarkdown": "", "isDeleted": true } ] } ], "comments": [ { "id": 686745, "author_name": "Nirjhar Roy", "author_url": "", "post_date": "2019-12-03T13:32:20.570000", "content": "

Oh man .. you are on a roll ! Pitch ,yaw too ..

", "votes": 3, "replies": [ { "id": 686749, "author_name": "DiegoJohnson", "author_url": "", "post_date": "2019-12-03T13:37:59.833000", "content": "

Hope you like it.

", "votes": 1, "replies": [] }, { "id": 686759, "author_name": "Nirjhar Roy", "author_url": "", "post_date": "2019-12-03T13:55:32.360000", "content": "

ofcourse , great find and great effort .

", "votes": 1, "replies": [] } ] }, { "id": 689285, "author_name": "Tehsheem Ismail", "author_url": "", "post_date": "2019-12-06T18:49:08.107000", "content": "

Very informative and helpful.

", "votes": 1, "replies": [ { "id": 689511, "author_name": "DiegoJohnson", "author_url": "", "post_date": "2019-12-07T01:52:19.683000", "content": "

thanks. I will keep updata, as I find useful information, you also can see my other topics.

", "votes": 1, "replies": [] } ] }, { "id": 688993, "author_name": "Solve for Fun", "author_url": "", "post_date": "2019-12-06T10:19:18.613000", "content": "

reference for me.

", "votes": 1, "replies": [] }, { "id": 687346, "author_name": "Ailurophile", "author_url": "", "post_date": "2019-12-04T09:20:09.270000", "content": "

Great Find\nVery Helpful\nThanks @diegojohnson

", "votes": 1, "replies": [ { "id": 687362, "author_name": "DiegoJohnson", "author_url": "", "post_date": "2019-12-04T09:39:27.043000", "content": "

Thanks😄

", "votes": 0, "replies": [] } ] }, { "id": 687282, "author_name": "Joey Chuang", "author_url": "", "post_date": "2019-12-04T07:10:18.170000", "content": "

Thank for your sharing. Can you kindly give us the link to this ppt?

", "votes": 1, "replies": [ { "id": 687285, "author_name": "DiegoJohnson", "author_url": "", "post_date": "2019-12-04T07:17:51.080000", "content": "

It's chinese, and I don't have ppt file.\nhttp://www.sohu.com/a/258872057_613224

", "votes": 1, "replies": [] } ] }, { "id": 686758, "author_name": "Jack Vial", "author_url": "", "post_date": "2019-12-03T13:53:16.817000", "content": "

Great find

", "votes": 2, "replies": [ { "id": 686763, "author_name": "DiegoJohnson", "author_url": "", "post_date": "2019-12-03T14:00:47.633000", "content": "

The source is chinese, so it's hard to find.😂

", "votes": 0, "replies": [] } ] }, { "id": 692335, "author_name": "Atul Anand {Jha}", "author_url": "", "post_date": "2019-12-11T06:40:38.060000", "content": "

Striking Post. You got a new follower ..

", "votes": 0, "replies": [ { "id": 692403, "author_name": "", "author_url": "", "post_date": "2019-12-11T08:15:52.263000", "content": "", "votes": 0, "replies": [] } ] } ], "raw_markdown_by_id": { "686742": "## Baidu Apollo\n\nBaidu Apollo has compared some different algorithms, and finally choose **YOLO** as main algorithm, modified **DarkNet** as Feature Extractor, because they have high requirement in speed.\n\n![1](http://file.elecfans.com/web1/M00/66/89/o4YBAFu9o3OAUdNsAAC81sWxRlE367.jpg)\n\n## They call this modified algorithm 'Multi task YOLO-3D'.\n\n![2](http://file.elecfans.com/web1/M00/66/89/o4YBAFu9o3SAHlJtAADazbyMU68795.jpg)\n\n## The parameters that be predicted:\n\n![3](http://file.elecfans.com/web1/M00/66/89/o4YBAFu9o3eAClnHAADNMJxXt00919.jpg)\n![4](http://file.elecfans.com/web1/M00/66/89/o4YBAFu9o3iAR2b4AADGD6L6W68535.jpg)\n\n## then they simplify the parameters, but we need to predict 'yaw, pitch, roll, x, y, z', and no D(length, width, height)\n\n![5](http://file.elecfans.com/web1/M00/66/89/o4YBAFu9o3mAPSYZAAD0_fFm_tw775.jpg)\n\n## It seems yaw is more important than pitch, roll.\n\n![6](http://file.elecfans.com/web1/M00/66/89/o4YBAFu9o3qAVgU5AADwEa89d3s458.jpg)\n\n## Train\n\n![7](http://file.elecfans.com/web1/M00/66/89/o4YBAFu9o3qAInKVAACk-xcwd70828.jpg)\n![8](http://file.elecfans.com/web1/M00/66/89/o4YBAFu9o3yAPT-HAAD5UysFgO0007.jpg)\n\n## We don't need the distance, but this picture is interesting.\n![9](http://file.elecfans.com/web1/M00/66/89/o4YBAFu9o36AJL0rAADst_qutg8528.jpg)\n\n## Whole pipeline:\n![10](http://file.elecfans.com/web1/M00/66/89/o4YBAFu9o4GAfNBjAACgfNjgGt8637.jpg)\n\n## It takes me a lot of time, if you think it's useful, please give me a upvote, thanks.\n\nThese are my other topics.\n[The coordinate system of this competition](https://www.kaggle.com/c/pku-autonomous-driving/discussion/123385)\n[Algorithm Selection for Beginner!](https://www.kaggle.com/c/pku-autonomous-driving/discussion/120015#latest-686515)\n[Understand Camera Intrinsic Parameters and X, Y, Z](https://www.kaggle.com/c/pku-autonomous-driving/discussion/120083#latest-687359)\n[Algorithm Selection in 6D Pose Estimation](https://www.kaggle.com/c/pku-autonomous-driving/discussion/120381)\n[How the Competition Organizer get train.csv !](https://www.kaggle.com/c/pku-autonomous-driving/discussion/120443)\n[Source of 6D Pose Estimation](https://www.kaggle.com/c/pku-autonomous-driving/discussion/120582)\n[A way to improve the accuracy of model](https://www.kaggle.com/c/pku-autonomous-driving/discussion/120653)\n[Object Detection in 20 Years](https://www.kaggle.com/c/pku-autonomous-driving/discussion/120710)\n\n\nMy notebook:\n[Centernet - Objects as Points](https://www.kaggle.com/diegojohnson/centernet-objects-as-points)\n[A Way to Regress Translation and Rotation](https://www.kaggle.com/diegojohnson/a-way-to-regress-translation-and-rotation?scriptVersionId=24677131)\n[Best Algorithm so far - Google's new paper in Nov](https://www.kaggle.com/diegojohnson/best-algorithm-so-far-google-s-new-paper-in-nov)\n[A better way to regress Yaw!](https://www.kaggle.com/diegojohnson/a-better-way-to-regress-yaw)\n[A clear view of car pose](https://www.kaggle.com/diegojohnson/a-clear-view-of-car-pose)\n[Dataset with preprocess](https://www.kaggle.com/c/pku-autonomous-driving/discussion/124480)", "686745": "Oh man .. you are on a roll ! Pitch ,yaw too ..", "689285": "Very informative and helpful.", "688993": "reference for me.", "687346": "Great Find\nVery Helpful\nThanks @diegojohnson ", "687282": "Thank for your sharing. Can you kindly give us the link to this ppt?", "686758": "Great find ", "692335": "Striking Post. You got a new follower .." } }