{ "id": 177110, "title": "Lyft Level5 Resources", "url": "/competitions/lyft-motion-prediction-autonomous-vehicles/discussion/177110", "author_name": "Peter", "post_date": "2020-08-24T21:49:51.258000", "votes": 57, "comment_count": 14, "views": 0, "content": "

Here are some useful links:

\n", "messages": [ { "id": 984147, "postDate": "2020-08-24T21:49:51.260Z", "content": "

Here are some useful links:

\n", "rawMarkdown": "Here are some useful links:\n\n- [Official site](https://self-driving.lyft.com/level5)\n- [L5Kit API Documentation](http://www.l5kit.org/)\n- [L5Kit Github Source](https://github.com/lyft/l5kit)\n- [L5Kit Tutorials](https://github.com/lyft/l5kit/blob/master/examples)\n- [Dataset Paper](https://arxiv.org/pdf/2006.14480.pdf)\n\n", "votes": 56 }, { "id": 984215, "postDate": "2020-08-25T00:30:21.147Z", "content": "

Thank you for share these well-informed resources @pestipeti๐Ÿ‘!

", "rawMarkdown": "Thank you for share these well-informed resources @pestipeti๐Ÿ‘!", "votes": 1 }, { "id": 984163, "postDate": "2020-08-24T22:34:07.483Z", "content": "

Hey @pestipeti, thank you a lot..as usual: early bird with a lot of useful and insightful resources. Thanks again!

", "rawMarkdown": "Hey @pestipeti, thank you a lot..as usual: early bird with a lot of useful and insightful resources. Thanks again!", "votes": 2 }, { "id": 997412, "postDate": "2020-09-04T03:16:40.983Z", "content": "

Hi,
\nI am using l5kit==1.0.6 and uploaded data as below.

\n
from l5kit.data import ChunkedDataset\ntrain_data = ChunkedDataset(\"/kaggle/input/lyft-motion-prediction-autonomous-vehicles/scenes/train.zarr\").open()\n
\n

But when I am trying to get \"traffic_light_faces\"

\n

train_data.traffic_light_faces[0]

\n

Its throwing exception. But I am using right version of l5kit. Please help me.

\n

AttributeError: 'ChunkedDataset' object has no attribute 'traffic_light_faces'

", "rawMarkdown": "Hi,\nI am using **l5kit==1.0.6** and uploaded data as below. \n\n```\nfrom l5kit.data import ChunkedDataset\ntrain_data = ChunkedDataset(\"/kaggle/input/lyft-motion-prediction-autonomous-vehicles/scenes/train.zarr\").open()\n```\nBut when I am trying to get \"traffic_light_faces\"\n\n`train_data.traffic_light_faces[0]`\n\nIts throwing exception. But I am using right version of l5kit. Please help me.\n\n**AttributeError: 'ChunkedDataset' object has no attribute 'traffic_light_faces'**", "replies": [ { "id": 997482, "postDate": "2020-09-04T04:31:19.577Z", "content": "

@pallaviroyal If you want direct access to the zarr data, try this:

\n
data = zarr.open_group(\"...YOUR_DIR.../train.zarr\", mode='r')\ndata['traffic_light_faces'][0]\n
", "rawMarkdown": "@pallaviroyal If you want direct access to the zarr data, try this:\n\n```\ndata = zarr.open_group(\"...YOUR_DIR.../train.zarr\", mode='r')\ndata['traffic_light_faces'][0]\n```", "votes": 2 }, { "id": 997852, "postDate": "2020-09-04T09:31:41.653Z", "content": "

@pestipeti Thank you! working fine. But have another question.
\nWhen I check data[\"scenes\"], It contains shape size 16265 and chunks size 10000.
\nHere ,

\n\n

Kindly help me.

", "rawMarkdown": "@pestipeti Thank you! working fine. But have another question.\nWhen I check **data[\"scenes\"]**, It contains **shape** size 16265 and **chunks** size 10000. \nHere , \n- what is mean by chunks? \n- per chunk how many records available? \n- what can be the overall available records in data[\"scenes\"]?\n- when I check, the unique of data[\"scenes\"][\"host\"], it just showing 4. but as per the dataset paper unique hosts must be 20. did I misunderstand something here? \n\nKindly help me.", "votes": 2 }, { "id": 997937, "postDate": "2020-09-04T10:55:44.493Z", "content": "

@pallaviroyal

\n", "rawMarkdown": "@pallaviroyal \n- I think chunks are zarr specific, you don't have to worry about it.\n- In the default dataset we have 16265 scenes, you should take a look at the full dataset too. There should be the missing scenes/hosts (I haven't checked it yet, though.)", "votes": 1 }, { "id": 997944, "postDate": "2020-09-04T11:02:08.637Z", "content": "

@pestipeti Sure, thanks for the clarification.

", "rawMarkdown": "@pestipeti Sure, thanks for the clarification." } ] }, { "id": 988964, "postDate": "2020-08-28T12:51:46.403Z", "content": "

Should I make predictions for target positions of the agent or future positions of the other traffic agents ?
\nand train.zarr set consists of 248 frames per scene while test set contains 100 , should I make predictions for final 50 frames per scene or for first 50 frames ?

", "rawMarkdown": "Should I make predictions for target positions of the agent or future positions of the other traffic agents ?\nand train.zarr set consists of 248 frames per scene while test set contains 100 , should I make predictions for final 50 frames per scene or for first 50 frames ?" }, { "id": 987708, "postDate": "2020-08-27T13:08:03.533Z", "content": "

Thanks for your priceless material saving my research time.

", "rawMarkdown": "Thanks for your priceless material saving my research time." }, { "id": 985942, "postDate": "2020-08-26T05:42:26.113Z", "content": "

Thanks @pestipeti for sharing. very useful and informative resources.

", "rawMarkdown": "Thanks @pestipeti for sharing. very useful and informative resources." }, { "id": 991137, "postDate": "2020-08-30T07:01:56.753Z", "rawMarkdown": "", "isDeleted": true }, { "id": 988872, "postDate": "2020-08-28T11:03:18.893Z", "content": "

very good resources! Thank you๐Ÿ˜€

", "rawMarkdown": "very good resources! Thank you๐Ÿ˜€" }, { "id": 987859, "postDate": "2020-08-27T15:21:32.800Z", "content": "

Thanks for sharing very helpful.

", "rawMarkdown": "Thanks for sharing very helpful." }, { "id": 987384, "postDate": "2020-08-27T07:57:07.187Z", "content": "

Thanks, these proved useful.๐Ÿ‘

", "rawMarkdown": "Thanks, these proved useful.๐Ÿ‘" } ], "comments": [ { "id": 984215, "author_name": "Beans", "author_url": "", "post_date": "2020-08-25T00:30:21.147000", "content": "

Thank you for share these well-informed resources @pestipeti๐Ÿ‘!

", "votes": 1, "replies": [] }, { "id": 984163, "author_name": "from coffee import *", "author_url": "", "post_date": "2020-08-24T22:34:07.483000", "content": "

Hey @pestipeti, thank you a lot..as usual: early bird with a lot of useful and insightful resources. Thanks again!

", "votes": 2, "replies": [] }, { "id": 997412, "author_name": "Pallavi Ramicetty", "author_url": "", "post_date": "2020-09-04T03:16:40.983000", "content": "

Hi,
\nI am using l5kit==1.0.6 and uploaded data as below.

\n
from l5kit.data import ChunkedDataset\ntrain_data = ChunkedDataset(\"/kaggle/input/lyft-motion-prediction-autonomous-vehicles/scenes/train.zarr\").open()\n
\n

But when I am trying to get \"traffic_light_faces\"

\n

train_data.traffic_light_faces[0]

\n

Its throwing exception. But I am using right version of l5kit. Please help me.

\n

AttributeError: 'ChunkedDataset' object has no attribute 'traffic_light_faces'

", "votes": 0, "replies": [ { "id": 997482, "author_name": "Peter", "author_url": "", "post_date": "2020-09-04T04:31:19.577000", "content": "

@pallaviroyal If you want direct access to the zarr data, try this:

\n
data = zarr.open_group(\"...YOUR_DIR.../train.zarr\", mode='r')\ndata['traffic_light_faces'][0]\n
", "votes": 2, "replies": [] }, { "id": 997852, "author_name": "Pallavi Ramicetty", "author_url": "", "post_date": "2020-09-04T09:31:41.653000", "content": "

@pestipeti Thank you! working fine. But have another question.
\nWhen I check data[\"scenes\"], It contains shape size 16265 and chunks size 10000.
\nHere ,

\n\n

Kindly help me.

", "votes": 2, "replies": [] }, { "id": 997937, "author_name": "Peter", "author_url": "", "post_date": "2020-09-04T10:55:44.493000", "content": "

@pallaviroyal

\n", "votes": 1, "replies": [] }, { "id": 997944, "author_name": "Pallavi Ramicetty", "author_url": "", "post_date": "2020-09-04T11:02:08.637000", "content": "

@pestipeti Sure, thanks for the clarification.

", "votes": 0, "replies": [] } ] }, { "id": 988964, "author_name": "th-blitz", "author_url": "", "post_date": "2020-08-28T12:51:46.403000", "content": "

Should I make predictions for target positions of the agent or future positions of the other traffic agents ?
\nand train.zarr set consists of 248 frames per scene while test set contains 100 , should I make predictions for final 50 frames per scene or for first 50 frames ?

", "votes": 0, "replies": [] }, { "id": 987708, "author_name": "Jeenwoo Park", "author_url": "", "post_date": "2020-08-27T13:08:03.533000", "content": "

Thanks for your priceless material saving my research time.

", "votes": 0, "replies": [] }, { "id": 985942, "author_name": "Prasanta Kundu", "author_url": "", "post_date": "2020-08-26T05:42:26.113000", "content": "

Thanks @pestipeti for sharing. very useful and informative resources.

", "votes": 0, "replies": [] }, { "id": 991137, "author_name": "", "author_url": "", "post_date": "2020-08-30T07:01:56.753000", "content": "", "votes": 0, "replies": [] }, { "id": 988872, "author_name": "Matrix", "author_url": "", "post_date": "2020-08-28T11:03:18.893000", "content": "

very good resources! Thank you๐Ÿ˜€

", "votes": 0, "replies": [] }, { "id": 987859, "author_name": "Bivek Subedi", "author_url": "", "post_date": "2020-08-27T15:21:32.800000", "content": "

Thanks for sharing very helpful.

", "votes": 0, "replies": [] }, { "id": 987384, "author_name": "Sushant Rai", "author_url": "", "post_date": "2020-08-27T07:57:07.187000", "content": "

Thanks, these proved useful.๐Ÿ‘

", "votes": 0, "replies": [] } ], "raw_markdown_by_id": { "984147": "Here are some useful links:\n\n- [Official site](https://self-driving.lyft.com/level5)\n- [L5Kit API Documentation](http://www.l5kit.org/)\n- [L5Kit Github Source](https://github.com/lyft/l5kit)\n- [L5Kit Tutorials](https://github.com/lyft/l5kit/blob/master/examples)\n- [Dataset Paper](https://arxiv.org/pdf/2006.14480.pdf)\n\n", "984215": "Thank you for share these well-informed resources @pestipeti๐Ÿ‘!", "984163": "Hey @pestipeti, thank you a lot..as usual: early bird with a lot of useful and insightful resources. Thanks again!", "997412": "Hi,\nI am using **l5kit==1.0.6** and uploaded data as below. \n\n```\nfrom l5kit.data import ChunkedDataset\ntrain_data = ChunkedDataset(\"/kaggle/input/lyft-motion-prediction-autonomous-vehicles/scenes/train.zarr\").open()\n```\nBut when I am trying to get \"traffic_light_faces\"\n\n`train_data.traffic_light_faces[0]`\n\nIts throwing exception. But I am using right version of l5kit. Please help me.\n\n**AttributeError: 'ChunkedDataset' object has no attribute 'traffic_light_faces'**", "988964": "Should I make predictions for target positions of the agent or future positions of the other traffic agents ?\nand train.zarr set consists of 248 frames per scene while test set contains 100 , should I make predictions for final 50 frames per scene or for first 50 frames ?", "987708": "Thanks for your priceless material saving my research time.", "985942": "Thanks @pestipeti for sharing. very useful and informative resources.", "991137": "", "988872": "very good resources! Thank you๐Ÿ˜€", "987859": "Thanks for sharing very helpful.", "987384": "Thanks, these proved useful.๐Ÿ‘" } }