{ "id": 109361, "title": "Official external data thread", "url": "/competitions/3d-object-detection-for-autonomous-vehicles/discussion/109361", "author_name": "Maggie", "post_date": "2019-09-18T16:59:56.420000", "votes": 10, "comment_count": 28, "views": 0, "content": "

Post links to your external data sources here.

", "messages": [ { "id": 629347, "postDate": "2019-09-18T16:59:56.420Z", "content": "

Post links to your external data sources here.

", "rawMarkdown": "Post links to your external data sources here.", "votes": 10 }, { "id": 629367, "postDate": "2019-09-18T17:21:11.487Z", "content": "

KITTI : http://www.cvlibs.net/datasets/kitti/\nnuScenes: https://www.nuscenes.org/overview

", "rawMarkdown": "KITTI : http://www.cvlibs.net/datasets/kitti/\nnuScenes: https://www.nuscenes.org/overview", "votes": 4, "replies": [ { "id": 629509, "postDate": "2019-09-18T21:17:58.470Z", "content": "

Sure, you can use them.

", "rawMarkdown": "Sure, you can use them. ", "votes": 4 }, { "id": 631281, "postDate": "2019-09-21T18:46:42.900Z", "content": "

Waymo Open dataset: https://waymo.com/open/about/

", "rawMarkdown": "Waymo Open dataset: https://waymo.com/open/about/", "votes": 4 }, { "id": 631402, "postDate": "2019-09-22T00:24:10.213Z", "content": "

Go for it :)

", "rawMarkdown": "Go for it :)", "votes": 2 }, { "id": 633913, "postDate": "2019-09-25T15:12:16.840Z", "content": "

One more! A2D2 from Audi https://www.audi-electronics-venture.com/aev/web/en/driving-dataset.html

", "rawMarkdown": "One more! A2D2 from Audi https://www.audi-electronics-venture.com/aev/web/en/driving-dataset.html", "votes": 2 }, { "id": 641656, "postDate": "2019-10-04T23:17:28.007Z", "content": "

Feel free to use it.

", "rawMarkdown": "Feel free to use it.", "votes": 1 } ] }, { "id": 665912, "postDate": "2019-11-05T14:24:01.437Z", "content": "

My pretrained models and dataset list:\nhttps://github.com/tensorflow/models/blob/master/research/object_detection/g3doc/detection_model_zoo.md\nhttps://github.com/facebookresearch/detectron2/blob/master/MODEL_ZOO.md\nhttps://github.com/open-mmlab/mmdetection/blob/master/docs/MODEL_ZOO.md\nhttps://github.com/Cadene/pretrained-models.pytorch\nhttps://www.nuscenes.org/download

", "rawMarkdown": "My pretrained models and dataset list:\nhttps://github.com/tensorflow/models/blob/master/research/object_detection/g3doc/detection_model_zoo.md\nhttps://github.com/facebookresearch/detectron2/blob/master/MODEL_ZOO.md\nhttps://github.com/open-mmlab/mmdetection/blob/master/docs/MODEL_ZOO.md\nhttps://github.com/Cadene/pretrained-models.pytorch\nhttps://www.nuscenes.org/download", "votes": 1, "replies": [ { "id": 665951, "postDate": "2019-11-05T15:14:13.493Z", "content": "

Feel free to use.

", "rawMarkdown": "Feel free to use.", "votes": 1 } ] }, { "id": 664946, "postDate": "2019-11-04T12:56:09.137Z", "content": "

I use pretrained models from: https://github.com/Cadene/pretrained-models.pytorch and https://github.com/open-mmlab/mmdetection.

", "rawMarkdown": "I use pretrained models from: https://github.com/Cadene/pretrained-models.pytorch and https://github.com/open-mmlab/mmdetection. ", "votes": 1, "replies": [ { "id": 665356, "postDate": "2019-11-04T23:11:39.163Z", "content": "

Feel free to use them.

", "rawMarkdown": "Feel free to use them.", "votes": 1 } ] }, { "id": 661919, "postDate": "2019-10-30T21:19:37.767Z", "content": "

I used the pretrained models from:\nhttps://github.com/lukemelas/EfficientNet-PyTorch\nhttps://github.com/pytorch/vision/tree/master/torchvision

\n\n

And i used the following images from wikipedia:

\n\n

https://en.wikipedia.org/wiki/Fire_engine\n\nHuachuca_City_Fire_-_2010-03-16_-_06.jpg\nFire_Company_Engine_4.jpg\nFire_Company_Tanker.jpg\nFire_Company_Tower.jpg\nFLFR_Ladder_13.JPG\nLFB_Pump_Ladder.jpg\nMercedes_Fire_truck,_Firedept_Antwerpen_Unit_A15.JPG\n

\n\n

https://en.wikipedia.org/wiki/Ambulance\n\nColumbus_Fire_Medic_7.JPG\nLAFD_ambulance.jpg\nR45-2.jpg\nAmbulance_sis.jpg\nMashpee_Mass._Ambulance_361_-_2007_Ford_E-450_Horton.jpg\nMedic_291.jpg\n

", "rawMarkdown": "I used the pretrained models from:\nhttps://github.com/lukemelas/EfficientNet-PyTorch\nhttps://github.com/pytorch/vision/tree/master/torchvision\n\nAnd i used the following images from wikipedia:\n\nhttps://en.wikipedia.org/wiki/Fire_engine\n```\nHuachuca_City_Fire_-_2010-03-16_-_06.jpg\nFire_Company_Engine_4.jpg\nFire_Company_Tanker.jpg\nFire_Company_Tower.jpg\nFLFR_Ladder_13.JPG\nLFB_Pump_Ladder.jpg\nMercedes_Fire_truck,_Firedept_Antwerpen_Unit_A15.JPG\n```\n\n\nhttps://en.wikipedia.org/wiki/Ambulance\n```\nColumbus_Fire_Medic_7.JPG\nLAFD_ambulance.jpg\nR45-2.jpg\nAmbulance_sis.jpg\nMashpee_Mass._Ambulance_361_-_2007_Ford_E-450_Horton.jpg\nMedic_291.jpg\n```\n\n", "votes": 1, "replies": [ { "id": 663958, "postDate": "2019-11-03T00:18:49.250Z", "content": "

Sure. Feel free to use it.

", "rawMarkdown": "Sure. Feel free to use it.", "votes": 1 } ] }, { "id": 655706, "postDate": "2019-10-23T12:00:12.073Z", "content": "

Hi ,I use the pertrain Unet Mode by Imagenet \nhttps://github.com/qubvel/segmentation_models.pytorch \nmodel = smp.Unet(\"resnet34\", classes=10, encoder_weights=\"imagenet\")

", "rawMarkdown": "Hi ,I use the pertrain Unet Mode by Imagenet \nhttps://github.com/qubvel/segmentation_models.pytorch \nmodel = smp.Unet(\"resnet34\", classes=10, encoder_weights=\"imagenet\")", "votes": 1 }, { "id": 651294, "postDate": "2019-10-17T09:44:16.853Z", "content": "

What about common pretraining datasets like ImageNet, COCO, CityScapes? Also, pretrained models from a Model Zoo like https://github.com/facebookresearch/detectron2

", "rawMarkdown": "What about common pretraining datasets like ImageNet, COCO, CityScapes? Also, pretrained models from a Model Zoo like https://github.com/facebookresearch/detectron2", "votes": 1, "replies": [ { "id": 651550, "postDate": "2019-10-17T15:41:55.527Z", "content": "

You can use them.

\n\n

But you need to post a link to the models to this thread.

\n\n

Say, you posted a link to detectron2 -> feel free to use them.

", "rawMarkdown": "You can use them.\n\nBut you need to post a link to the models to this thread.\n\nSay, you posted a link to detectron2 -> feel free to use them.", "votes": 1 }, { "id": 657800, "postDate": "2019-10-25T12:36:25.433Z", "content": "

Hi ,can I use the pertrain Unet Mode by Imagenet \nhttps://github.com/qubvel/segmentation_models.pytorch \nmodel = smp.Unet(\"resnet34\", classes=10, encoder_weights=\"imagenet\")

", "rawMarkdown": "Hi ,can I use the pertrain Unet Mode by Imagenet \nhttps://github.com/qubvel/segmentation_models.pytorch \nmodel = smp.Unet(\"resnet34\", classes=10, encoder_weights=\"imagenet\")" }, { "id": 665357, "postDate": "2019-11-04T23:12:09.030Z", "content": "

Feel free to use all pre-trained models from that repo.

", "rawMarkdown": "Feel free to use all pre-trained models from that repo." } ] }, { "id": 654016, "postDate": "2019-10-21T10:07:53.180Z", "content": "

Argoverse https://www.argoverse.org\nbdd100k https://bdd-data.berkeley.edu/\nFisher Yu's GTA dataset http://dl.yf.io/bdd-data/3d-vehicle-tracking/ https://github.com/ucbdrive/3d-vehicle-tracking/blob/master/3d-tracking/loader/download.py#L11

\n\n

can we use data we created with Microsoft AirSim? https://microsoft.github.io/AirSim/

", "rawMarkdown": "Argoverse https://www.argoverse.org\nbdd100k https://bdd-data.berkeley.edu/\nFisher Yu's GTA dataset http://dl.yf.io/bdd-data/3d-vehicle-tracking/ https://github.com/ucbdrive/3d-vehicle-tracking/blob/master/3d-tracking/loader/download.py#L11 \n\ncan we use data we created with Microsoft AirSim? https://microsoft.github.io/AirSim/ ", "votes": 2, "replies": [ { "id": 654039, "postDate": "2019-10-21T10:55:22.877Z", "content": "
\n

can we use data we created with Microsoft AirSim?

\n
\n\n

Probably not, the competition rules allow only publicly available datasets to be used.

", "rawMarkdown": "> can we use data we created with Microsoft AirSim? \n\nProbably not, the competition rules allow only publicly available datasets to be used.", "votes": 1 }, { "id": 655265, "postDate": "2019-10-22T21:26:52.367Z", "content": "

Let's say:

\n\n

Argoverse, bdd100k, Fisher Yu's GTA dataset are ok to use.

\n\n

Microsoft AirSim - I will look more at it, but for now, let's not use it.

", "rawMarkdown": "Let's say:\n\nArgoverse, bdd100k, Fisher Yu's GTA dataset are ok to use.\n\nMicrosoft AirSim - I will look more at it, but for now, let's not use it." } ] }, { "id": 668202, "postDate": "2019-11-08T05:42:44.773Z", "content": "

https://www.openstreetmap.org/ for Palo Alto area

", "rawMarkdown": "https://www.openstreetmap.org/ for Palo Alto area" }, { "id": 666177, "postDate": "2019-11-05T20:47:20.560Z", "content": "

I may use some of the following public external datasets for pretraining:

\n\n

KITTI : http://www.cvlibs.net/datasets/kitti/\nnuScenes: https://www.nuscenes.org/overview\nWaymo: https://waymo.com/open/about/\nArgoverse: https://www.argoverse.org\nBDD100k: https://bdd-data.berkeley.edu/\nGTA: http://dl.yf.io/bdd-data/3d-vehicle-tracking/ https://github.com/ucbdrive/3d-vehicle-tracking/blob/master/3d-tracking/loader/download.py#L11

\n\n

I also may or may not use some pretrained backbone models from:

\n\n

https://github.com/facebookresearch/detectron2/blob/master/MODEL_ZOO.md\nhttps://github.com/pytorch/vision\nhttps://github.com/qubvel/segmentation_models.pytorch\nhttps://github.com/open-mmlab/mmdetection/blob/master/docs/MODEL_ZOO.md\nhttps://github.com/rwightman/pytorch-image-models\nhttps://github.com/Cadene/pretrained-models.pytorch\nhttps://github.com/lukemelas/EfficientNet-PyTorch

", "rawMarkdown": "I may use some of the following public external datasets for pretraining:\n\nKITTI : http://www.cvlibs.net/datasets/kitti/\nnuScenes: https://www.nuscenes.org/overview\nWaymo: https://waymo.com/open/about/\nArgoverse: https://www.argoverse.org\nBDD100k: https://bdd-data.berkeley.edu/\nGTA: http://dl.yf.io/bdd-data/3d-vehicle-tracking/ https://github.com/ucbdrive/3d-vehicle-tracking/blob/master/3d-tracking/loader/download.py#L11\n\nI also may or may not use some pretrained backbone models from:\n\nhttps://github.com/facebookresearch/detectron2/blob/master/MODEL_ZOO.md\nhttps://github.com/pytorch/vision\nhttps://github.com/qubvel/segmentation_models.pytorch\nhttps://github.com/open-mmlab/mmdetection/blob/master/docs/MODEL_ZOO.md\nhttps://github.com/rwightman/pytorch-image-models\nhttps://github.com/Cadene/pretrained-models.pytorch\nhttps://github.com/lukemelas/EfficientNet-PyTorch\n", "replies": [ { "id": 667001, "postDate": "2019-11-06T17:50:22.810Z", "content": "

Feel free to use it.

", "rawMarkdown": "Feel free to use it." }, { "id": 668608, "postDate": "2019-11-08T16:04:33.867Z", "content": "

https://github.com/NVIDIA/semantic-segmentation

", "rawMarkdown": "https://github.com/NVIDIA/semantic-segmentation" } ] }, { "id": 666126, "postDate": "2019-11-05T18:50:36.270Z", "content": "

https://github.com/traveller59/second.pytorch

", "rawMarkdown": "https://github.com/traveller59/second.pytorch" }, { "id": 665900, "postDate": "2019-11-05T14:08:08.973Z", "content": "

https://github.com/rwightman/pytorch-image-models/tree/master/timm/models\nhttps://github.com/pytorch/vision

", "rawMarkdown": "https://github.com/rwightman/pytorch-image-models/tree/master/timm/models\nhttps://github.com/pytorch/vision" }, { "id": 648820, "postDate": "2019-10-14T16:33:40.357Z", "content": "

Can we use WikiCommons Images from Wikipedia sites? For example: https://en.wikipedia.org/wiki/Police_vehicles_in_the_United_States_and_Canada

", "rawMarkdown": "Can we use WikiCommons Images from Wikipedia sites? For example: https://en.wikipedia.org/wiki/Police_vehicles_in_the_United_States_and_Canada" }, { "id": 641072, "postDate": "2019-10-04T12:23:59.373Z", "rawMarkdown": "", "isDeleted": true } ], "comments": [ { "id": 629367, "author_name": "Rishabh Agrahari", "author_url": "", "post_date": "2019-09-18T17:21:11.487000", "content": "

KITTI : http://www.cvlibs.net/datasets/kitti/\nnuScenes: https://www.nuscenes.org/overview

", "votes": 4, "replies": [ { "id": 629509, "author_name": "Vladimir Iglovikov", "author_url": "", "post_date": "2019-09-18T21:17:58.470000", "content": "

Sure, you can use them.

", "votes": 4, "replies": [] }, { "id": 631281, "author_name": "Rishabh Agrahari", "author_url": "", "post_date": "2019-09-21T18:46:42.900000", "content": "

Waymo Open dataset: https://waymo.com/open/about/

", "votes": 4, "replies": [] }, { "id": 631402, "author_name": "Vladimir Iglovikov", "author_url": "", "post_date": "2019-09-22T00:24:10.213000", "content": "

Go for it :)

", "votes": 2, "replies": [] }, { "id": 633913, "author_name": "Willem Prins", "author_url": "", "post_date": "2019-09-25T15:12:16.840000", "content": "

One more! A2D2 from Audi https://www.audi-electronics-venture.com/aev/web/en/driving-dataset.html

", "votes": 2, "replies": [] }, { "id": 641656, "author_name": "Vladimir Iglovikov", "author_url": "", "post_date": "2019-10-04T23:17:28.007000", "content": "

Feel free to use it.

", "votes": 1, "replies": [] } ] }, { "id": 665912, "author_name": "Kyle Lee", "author_url": "", "post_date": "2019-11-05T14:24:01.437000", "content": "

My pretrained models and dataset list:\nhttps://github.com/tensorflow/models/blob/master/research/object_detection/g3doc/detection_model_zoo.md\nhttps://github.com/facebookresearch/detectron2/blob/master/MODEL_ZOO.md\nhttps://github.com/open-mmlab/mmdetection/blob/master/docs/MODEL_ZOO.md\nhttps://github.com/Cadene/pretrained-models.pytorch\nhttps://www.nuscenes.org/download

", "votes": 1, "replies": [ { "id": 665951, "author_name": "Vladimir Iglovikov", "author_url": "", "post_date": "2019-11-05T15:14:13.493000", "content": "

Feel free to use.

", "votes": 1, "replies": [] } ] }, { "id": 664946, "author_name": "Ivan Sosin", "author_url": "", "post_date": "2019-11-04T12:56:09.137000", "content": "

I use pretrained models from: https://github.com/Cadene/pretrained-models.pytorch and https://github.com/open-mmlab/mmdetection.

", "votes": 1, "replies": [ { "id": 665356, "author_name": "Vladimir Iglovikov", "author_url": "", "post_date": "2019-11-04T23:11:39.163000", "content": "

Feel free to use them.

", "votes": 1, "replies": [] } ] }, { "id": 661919, "author_name": "Marek Wyborski", "author_url": "", "post_date": "2019-10-30T21:19:37.767000", "content": "

I used the pretrained models from:\nhttps://github.com/lukemelas/EfficientNet-PyTorch\nhttps://github.com/pytorch/vision/tree/master/torchvision

\n\n

And i used the following images from wikipedia:

\n\n

https://en.wikipedia.org/wiki/Fire_engine\n\nHuachuca_City_Fire_-_2010-03-16_-_06.jpg\nFire_Company_Engine_4.jpg\nFire_Company_Tanker.jpg\nFire_Company_Tower.jpg\nFLFR_Ladder_13.JPG\nLFB_Pump_Ladder.jpg\nMercedes_Fire_truck,_Firedept_Antwerpen_Unit_A15.JPG\n

\n\n

https://en.wikipedia.org/wiki/Ambulance\n\nColumbus_Fire_Medic_7.JPG\nLAFD_ambulance.jpg\nR45-2.jpg\nAmbulance_sis.jpg\nMashpee_Mass._Ambulance_361_-_2007_Ford_E-450_Horton.jpg\nMedic_291.jpg\n

", "votes": 1, "replies": [ { "id": 663958, "author_name": "Vladimir Iglovikov", "author_url": "", "post_date": "2019-11-03T00:18:49.250000", "content": "

Sure. Feel free to use it.

", "votes": 1, "replies": [] } ] }, { "id": 655706, "author_name": "AAA", "author_url": "", "post_date": "2019-10-23T12:00:12.073000", "content": "

Hi ,I use the pertrain Unet Mode by Imagenet \nhttps://github.com/qubvel/segmentation_models.pytorch \nmodel = smp.Unet(\"resnet34\", classes=10, encoder_weights=\"imagenet\")

", "votes": 1, "replies": [] }, { "id": 651294, "author_name": "Tobias Fischer", "author_url": "", "post_date": "2019-10-17T09:44:16.853000", "content": "

What about common pretraining datasets like ImageNet, COCO, CityScapes? Also, pretrained models from a Model Zoo like https://github.com/facebookresearch/detectron2

", "votes": 1, "replies": [ { "id": 651550, "author_name": "Vladimir Iglovikov", "author_url": "", "post_date": "2019-10-17T15:41:55.527000", "content": "

You can use them.

\n\n

But you need to post a link to the models to this thread.

\n\n

Say, you posted a link to detectron2 -> feel free to use them.

", "votes": 1, "replies": [] }, { "id": 657800, "author_name": "AAA", "author_url": "", "post_date": "2019-10-25T12:36:25.433000", "content": "

Hi ,can I use the pertrain Unet Mode by Imagenet \nhttps://github.com/qubvel/segmentation_models.pytorch \nmodel = smp.Unet(\"resnet34\", classes=10, encoder_weights=\"imagenet\")

", "votes": 0, "replies": [] }, { "id": 665357, "author_name": "Vladimir Iglovikov", "author_url": "", "post_date": "2019-11-04T23:12:09.030000", "content": "

Feel free to use all pre-trained models from that repo.

", "votes": 0, "replies": [] } ] }, { "id": 654016, "author_name": "oarph", "author_url": "", "post_date": "2019-10-21T10:07:53.180000", "content": "

Argoverse https://www.argoverse.org\nbdd100k https://bdd-data.berkeley.edu/\nFisher Yu's GTA dataset http://dl.yf.io/bdd-data/3d-vehicle-tracking/ https://github.com/ucbdrive/3d-vehicle-tracking/blob/master/3d-tracking/loader/download.py#L11

\n\n

can we use data we created with Microsoft AirSim? https://microsoft.github.io/AirSim/

", "votes": 2, "replies": [ { "id": 654039, "author_name": "Rishabh Agrahari", "author_url": "", "post_date": "2019-10-21T10:55:22.877000", "content": "
\n

can we use data we created with Microsoft AirSim?

\n
\n\n

Probably not, the competition rules allow only publicly available datasets to be used.

", "votes": 1, "replies": [] }, { "id": 655265, "author_name": "Vladimir Iglovikov", "author_url": "", "post_date": "2019-10-22T21:26:52.367000", "content": "

Let's say:

\n\n

Argoverse, bdd100k, Fisher Yu's GTA dataset are ok to use.

\n\n

Microsoft AirSim - I will look more at it, but for now, let's not use it.

", "votes": 0, "replies": [] } ] }, { "id": 668202, "author_name": "Arizona Dad", "author_url": "", "post_date": "2019-11-08T05:42:44.773000", "content": "

https://www.openstreetmap.org/ for Palo Alto area

", "votes": 0, "replies": [] }, { "id": 666177, "author_name": "aagapi", "author_url": "", "post_date": "2019-11-05T20:47:20.560000", "content": "

I may use some of the following public external datasets for pretraining:

\n\n

KITTI : http://www.cvlibs.net/datasets/kitti/\nnuScenes: https://www.nuscenes.org/overview\nWaymo: https://waymo.com/open/about/\nArgoverse: https://www.argoverse.org\nBDD100k: https://bdd-data.berkeley.edu/\nGTA: http://dl.yf.io/bdd-data/3d-vehicle-tracking/ https://github.com/ucbdrive/3d-vehicle-tracking/blob/master/3d-tracking/loader/download.py#L11

\n\n

I also may or may not use some pretrained backbone models from:

\n\n

https://github.com/facebookresearch/detectron2/blob/master/MODEL_ZOO.md\nhttps://github.com/pytorch/vision\nhttps://github.com/qubvel/segmentation_models.pytorch\nhttps://github.com/open-mmlab/mmdetection/blob/master/docs/MODEL_ZOO.md\nhttps://github.com/rwightman/pytorch-image-models\nhttps://github.com/Cadene/pretrained-models.pytorch\nhttps://github.com/lukemelas/EfficientNet-PyTorch

", "votes": 0, "replies": [ { "id": 667001, "author_name": "Vladimir Iglovikov", "author_url": "", "post_date": "2019-11-06T17:50:22.810000", "content": "

Feel free to use it.

", "votes": 0, "replies": [] }, { "id": 668608, "author_name": "Kishore M", "author_url": "", "post_date": "2019-11-08T16:04:33.867000", "content": "

https://github.com/NVIDIA/semantic-segmentation

", "votes": 0, "replies": [] } ] }, { "id": 666126, "author_name": "Mikhail Edukov", "author_url": "", "post_date": "2019-11-05T18:50:36.270000", "content": "

https://github.com/traveller59/second.pytorch

", "votes": 0, "replies": [] }, { "id": 665900, "author_name": "See--", "author_url": "", "post_date": "2019-11-05T14:08:08.973000", "content": "

https://github.com/rwightman/pytorch-image-models/tree/master/timm/models\nhttps://github.com/pytorch/vision

", "votes": 0, "replies": [] }, { "id": 648820, "author_name": "Marek Wyborski", "author_url": "", "post_date": "2019-10-14T16:33:40.357000", "content": "

Can we use WikiCommons Images from Wikipedia sites? For example: https://en.wikipedia.org/wiki/Police_vehicles_in_the_United_States_and_Canada

", "votes": 0, "replies": [] }, { "id": 641072, "author_name": "", "author_url": "", "post_date": "2019-10-04T12:23:59.373000", "content": "", "votes": 0, "replies": [] } ], "raw_markdown_by_id": { "629347": "Post links to your external data sources here.", "629367": "KITTI : http://www.cvlibs.net/datasets/kitti/\nnuScenes: https://www.nuscenes.org/overview", "665912": "My pretrained models and dataset list:\nhttps://github.com/tensorflow/models/blob/master/research/object_detection/g3doc/detection_model_zoo.md\nhttps://github.com/facebookresearch/detectron2/blob/master/MODEL_ZOO.md\nhttps://github.com/open-mmlab/mmdetection/blob/master/docs/MODEL_ZOO.md\nhttps://github.com/Cadene/pretrained-models.pytorch\nhttps://www.nuscenes.org/download", "664946": "I use pretrained models from: https://github.com/Cadene/pretrained-models.pytorch and https://github.com/open-mmlab/mmdetection. ", "661919": "I used the pretrained models from:\nhttps://github.com/lukemelas/EfficientNet-PyTorch\nhttps://github.com/pytorch/vision/tree/master/torchvision\n\nAnd i used the following images from wikipedia:\n\nhttps://en.wikipedia.org/wiki/Fire_engine\n```\nHuachuca_City_Fire_-_2010-03-16_-_06.jpg\nFire_Company_Engine_4.jpg\nFire_Company_Tanker.jpg\nFire_Company_Tower.jpg\nFLFR_Ladder_13.JPG\nLFB_Pump_Ladder.jpg\nMercedes_Fire_truck,_Firedept_Antwerpen_Unit_A15.JPG\n```\n\n\nhttps://en.wikipedia.org/wiki/Ambulance\n```\nColumbus_Fire_Medic_7.JPG\nLAFD_ambulance.jpg\nR45-2.jpg\nAmbulance_sis.jpg\nMashpee_Mass._Ambulance_361_-_2007_Ford_E-450_Horton.jpg\nMedic_291.jpg\n```\n\n", "655706": "Hi ,I use the pertrain Unet Mode by Imagenet \nhttps://github.com/qubvel/segmentation_models.pytorch \nmodel = smp.Unet(\"resnet34\", classes=10, encoder_weights=\"imagenet\")", "651294": "What about common pretraining datasets like ImageNet, COCO, CityScapes? Also, pretrained models from a Model Zoo like https://github.com/facebookresearch/detectron2", "654016": "Argoverse https://www.argoverse.org\nbdd100k https://bdd-data.berkeley.edu/\nFisher Yu's GTA dataset http://dl.yf.io/bdd-data/3d-vehicle-tracking/ https://github.com/ucbdrive/3d-vehicle-tracking/blob/master/3d-tracking/loader/download.py#L11 \n\ncan we use data we created with Microsoft AirSim? https://microsoft.github.io/AirSim/ ", "668202": "https://www.openstreetmap.org/ for Palo Alto area", "666177": "I may use some of the following public external datasets for pretraining:\n\nKITTI : http://www.cvlibs.net/datasets/kitti/\nnuScenes: https://www.nuscenes.org/overview\nWaymo: https://waymo.com/open/about/\nArgoverse: https://www.argoverse.org\nBDD100k: https://bdd-data.berkeley.edu/\nGTA: http://dl.yf.io/bdd-data/3d-vehicle-tracking/ https://github.com/ucbdrive/3d-vehicle-tracking/blob/master/3d-tracking/loader/download.py#L11\n\nI also may or may not use some pretrained backbone models from:\n\nhttps://github.com/facebookresearch/detectron2/blob/master/MODEL_ZOO.md\nhttps://github.com/pytorch/vision\nhttps://github.com/qubvel/segmentation_models.pytorch\nhttps://github.com/open-mmlab/mmdetection/blob/master/docs/MODEL_ZOO.md\nhttps://github.com/rwightman/pytorch-image-models\nhttps://github.com/Cadene/pretrained-models.pytorch\nhttps://github.com/lukemelas/EfficientNet-PyTorch\n", "666126": "https://github.com/traveller59/second.pytorch", "665900": "https://github.com/rwightman/pytorch-image-models/tree/master/timm/models\nhttps://github.com/pytorch/vision", "648820": "Can we use WikiCommons Images from Wikipedia sites? For example: https://en.wikipedia.org/wiki/Police_vehicles_in_the_United_States_and_Canada", "641072": "" } }