Our datasets are hosted on a publicly accessible S3 bucket. You can use the aws-cli to download individual objects or the whole dataset.
You can list all datasets and parts thereof like so:
aws s3 ls --summarize --human-readable --recursive s3://public-datasets/ --endpoint=https://web.s3.wisc.edu --no-sign-request
To only list data associated with a single dataset replace the URI above with one that matches the dataset prefix, e.g by using s3://public-datasets/quanta-vision/sequences.
To download a specific object (where object-key is eg quanta-vision/sequences/README.md) you can use the following command:
aws s3api get-object --bucket public-datasets --key <OBJECT-KEY> --endpoint=https://web.s3.wisc.edu --no-sign-request <DOWNLOAD-PATH>
Finally, here's an example script which will download and unzip the whole quanta-vision/sequences dataset (warning ~2.4TB). You can use the same script with a different DATASET_PREFIX to download other datasets or subparts thereof:
#!/usr/bin/env bash
# Directory to download data to
DOWNLOAD_DIR=downloads/
DATASET_PREFIX=quanta-vision/sequences
# Clone all data from S3
aws s3 sync s3://public-datasets/$DATASET_PREFIX $DOWNLOAD_DIR --endpoint=https://web.s3.wisc.edu --no-sign-request
# Extract all zips in their CWD
for zip in $(find $DOWNLOAD_DIR -type f -name *.zip); do 7z x $zip -o$(dirname $zip) && rm -f $zip; done
Note: Some archives use LZMA for higher compression ratios, you can use the 7z cli to unzip these (as above), but the unzip command might not work.
Note: If you are on UW-Madison wifi or connected to the campus VPN, downloads will be much faster.
Below we include folder-wise descriptions (of directories under sequences), paper(s) associated with the folder and hot-pixel masks per sequence. These real-world sequences were captured using the passive single photon cameras, high speed cameras, or other specialized cameras (event/low light cameras).
anycam: sequences associated with Sundar et al., ICCV 2023. All sequences were captured at 96.8 kHz. Associated hot-pixel mask ishot_pixel_mask/SwissSPAD_ddr3_mode.npyfor sequences captured by the SPAD with no color filter array andhot_pixel_mask/colorSPAD_continuous_stream.npyfor the rest. Seearguments.jsonin each folder that contains abinary.npyfile for discerning which is which. Alternatively, the mean-frame video gives it away (ones that use a CFA have a conspicuous mosaic pattern). When using the color-filter array captured sequences, we impute out the pixels corresponding to "R", "G", and "B" filters; these are a minority and make up just 6.25% of the overall pixel count.color: sequences associated with Ma et al., SIGGRAPH 2023. Sequences were captured at 16 kHz (unless annotated otherwise) and use the hotpixel mask inhot_pixel_mask/colorSPAD_continuous_stream.npy. Seecolor_filter_array/rgbw_oh_bn_color_ss2_padded.tiffor a specification of the random RGBW CFA pattern.pano: sequences associated with Jungerman et al., ICCV 2023. Sequences were captured at 96.8 kHz. Hot-pixel mask specified byhot_pixel_mask/colorSPAD_continuous_stream.npy. We use the color-filter array pattern with 93.75 % white or clear pixels and impute out the photon-cube locations associated with a "R", "G", or "B" photon detection.photoev: sequences associated with Sundar et al., CVPR 2024. All sequences were captured at 96.8 kHz. Hot pixel masks arehot_pixel_mask/new_graySPAD_continuous_stream.npyfor sequences captured by the SPAD with no CFAs andhot_pixel_mask/colorSPAD_continuous_stream.npyotherwise.qbp: sequences associated with Ma et al., SIGGRAPH 2020. Hotpixel mask for all sequences ishot_pixel_mask/SwissSPAD_ddr3_mode.npy. Sequences captured at 10--16 kHz.vision: sequences associated with Ma et al., WACV 2023. Hotpixel mask for all sequences ishot_pixel_mask/SwissSPAD_continuous_stream.npy. Most sequences were captured at 10--16 kHz.
See Detailed Folder Structure
ROOT: quanta_vision/sequences
├── (ZIP 29.9G) 📁 qbp
├── (ZIP 927.4K) 📁 masks
├── 📁 anycam
│ ├── (ZIP #1/3 98.8G) balloon_burst_17th_Dec_2022, bubble_machine_17th_Dec_2022, capitol_24th_Feb_2023, casino_roulette_10th_Feb_2023, confetti_popper_17th_Dec_2022, eye_track_17th_Dec_2022, falling_dice_9th_Dec_2022, falling_dice_11th_Dec_2022, jack-in-the-box_17th_Dec_2022, measuring_tape_17th_Dec_2022
│ ├── (ZIP #2/3 94.3G) newton_cradle_8th_Feb_2023, party_popper_17th_Dec_2022, pedestrian_24th_Feb_2023, ramanujam_24th_Feb_2023, sanity, tabletop_24th_Feb_2023, traffic_10th_Feb_2023, vanvleck_24th_Feb_2023
│ └── (ZIP #3/3 101.4G) vertical_wheel_10th_Feb_2023, vertical_wheel_17th_Dec_2022, vertical_wheel_colorSPAD_10th_Feb_2023, water_meniscus_17th_Dec_2022, falling_dice.mp4
├── 📁 color
│ ├── (ZIP #1/6 90.4G) 1221_May_8th, 1240_May_8th, 1240_backstage_May_8th, 1240_gray_panel_May_8th, 1325_May_8th, HDR_April_27th, HDR_white_vase_10th_November, HDR_white_vase_19th_October, LED_balloon_May_3rd
│ ├── (ZIP #2/6 108.6G) LED_balloons_Mat_13th, all_dark, all_dark_25th_September, all_white, balloon_burst_April_27th, bouncy_balls_July_12th, bouncy_balls_July_27th, bubbles_April_28th, casino_roulette_July_12th
│ ├── (ZIP #3/6 99.7G) casino_roulette_July_27th, chair_May_28th, cloth_April_26th, color_chart_April_26th, colored_dice_July_27th, dartboard_May_25th, darts_April_26th, dice_July_8th, dry_run, dry_run_April_14th, dry_run_April_14th_8pm, entrance_HDR_July_27th, entrance_May_8th
│ ├── (ZIP #4/6 84.5G) entrance_May_16th, entrance_May_27th, entrance_May_30th, feathers_April_27th, feathers_May_30th, fence_structure_June_8th, fence_structure_May_30th
│ ├── (ZIP #5/6 94.6G) front_entrance_May_16th, fruits_May_30th, grafitti_elephant_20th_October, hdr_entrance_July_21st, hdr_entrance_table_July_21st, jack-in-the-box_June_8th, jack-in-the-box_May_30th, lighter_April_21st, lighter_May_3rd, potted_plant_1309_May_27th, potted_plant_June_8th
│ └── (ZIP #6/6 93.5G) tabletop_April_20th, toy_fence_June_1st, vase_HDR_5th_April_2023, vase_HDR_Sept_13th, vertical_wheel_July_11th, vertical_wheel_July_15th, waveform_LED_17th_April_2023, waving_cloth_May_30th, rgbw_oh_bn_color_ss2_padded.tif
├── 📁 pano
│ ├── (ZIP #1/2 82.4G) cs6floorlounge, vanvleck
│ └── (ZIP #2/2 92.0G) vanvleck2
├── 📁 photoev
│ ├── (ZIP #1/2 97.4G) blender_1st_Sept, blender_almonds_1st_Sept, blender_almonds_take_2_1st_Sept, blender_almonds_take_3_1st_Sept, darts_22nd_Sept, darts_26th_Sept_ambient, darts_26th_Sept_dark, darts_26th_Sept_dark_1lux, darts_26th_Sept_dark_2lux_2023-09-26--16-56-48, darts_26th_Sept_dark_5lux, darts_low_light_2_22nd_Sept, darts_low_light_3_22nd_Sept, darts_low_light_4_22nd_Sept, darts_low_light_22nd_Sept, drill_1st_Sept, drill_take_2_1st_Sept, drill_take_3_1st_Sept, dslr_shutter, flag_6th_floor_13th_Sept, flag_6th_floor_13th_Sept_take2, iphone_lock_screen, iphone_lock_screen_20_per, iphone_lock_screen_80_per, iphone_screen_20_per, iphone_screen_20_per_2023-11-10--19-00-49, iphone_screen_80_per, leaf_blower_1st_Sept, lighter_1st_Sept, lighter_take_2_1st_Sept
│ └── (ZIP #2/2 92.3G) lighter_take_3_1st_Sept, phone, phone_screen, prophesee, slingshot_1st_Sept, slingshot_13th_Nov_2023-11-13--14-32-28, slingshot_13th_Nov_2023-11-13--14-38-03, slingshot_13th_Nov_2023-11-13--14-41-33, slingshot_13th_Nov_2023-11-13--14-44-55, slingshot_13th_Nov_prophesee, stress_ball_1st_Sept, stress_ball_take_2_1st_Sept, stressball_29th_Sept, stressball_29th_Sept_12mm_prophesee, stressball_29th_Sept_16mm_infinicam, stressball_29th_Sept_2023-09-29--15-11-44, tennis_27th_Sept_75mm_2023-09-27--17-12-32, tennis_27th_Sept_75mm_2023-09-27--17-13-13, tennis_27th_Sept_75mm_2023-09-27--17-14-18, tennis_50mm_27th_Sept_rear_2023-09-27--17-58-43, tennis_50mm_27th_Sept_rear_2023-09-27--17-59-23, tennis_50mm_27th_Sept_rear_2023-09-27--17-59-58, tennis_50mm_27th_Sept_rear_2023-09-27--18-00-44, tennis_100mm_27th_Sept_2023-09-27--17-37-52, tennis_100mm_27th_Sept_2023-09-27--17-39-05, tennis_prophesee, traffic_8pm_27th_Sept_2023-09-27--20-04-33, traffic_8pm_27th_Sept_2023-09-27--20-11-42, traffic_8pm_27th_Sept_2023-09-27--20-14-31, traffic_8pm_27th_Sept_2023-09-27--20-17-33, traffic_8pm_27th_Sept_prophesee
├── 📁 vision
│ ├── (ZIP #1/12 97.7G) 0505-bicycle-1, 0505-bicycle-2, 0505-bicycle-3, 0505-bicycle-4, 0505-bicycle-5, 0505-face-1, 0505-face-2, 0505-face-3, 0505-face-4, 0525-newton-1, 0525-newton-2, 0525-newton-3, 0525-newton-4, 0525-newton-5, 0527-train-bright, 0527-train-switch, 0528-pendulum-1, 0528-pendulum-2, 0528-pendulum-3, 0528-train-dark-1, 0528-train-dark-2, 0528-train-switch-1, 0528-train-switch-2, 0531-spinner-1, 0531-spinner-2
│ ├── (ZIP #2/12 98.4G) 0531-spinner-3, 0602-street, 0604-actions-1, 0604-actions-2, 0604-actions-3, 0604-ball-1, 0604-ball-2, 0604-ball-3, 0604-chair-0, 0604-chair-1, 0604-chair-2, 0604-chair-3, 0604-face-0, 0604-face-1, 0604-face-2, 0604-face-3, 0604-jump-1, 0604-jump-2, 0604-jump-3, 0604-runwalk-1, 0604-runwalk-2, 0604-runwalk-3, 0604-throwdrink-0, 0604-throwdrink-1, 0604-throwdrink-2, 0604-throwdrink-3, 0604-walk-1, 0604-walk-2, 0604-walk-3, 0604-walk-4, 0604-walk-5, 0608-street-1
│ ├── (ZIP #3/12 99.5G) 0608-street-2, 0608-street-3, 0609-handheld-1, 0609-handheld-2, 0609-handheld-3, 0614-calib-1, 0614-calib-2, 0614-calib-3, 0614-calib-4, 0614-calib-5, 0702-moving-bike-dark-1, 0702-moving-drive, 0702-moving-ocr-1, 0702-moving-ocr-2, 0702-moving-ocr-3, 0702-moving-walktoward-1, 0702-moving-walktoward-dark-1, 0702-moving-walktoward-dark-2, 0702-static-bike-1, 0702-static-bike-2, 0702-static-bike-dark-1, 0702-static-jump-ddark-1, 0702-static-run-1
│ ├── (ZIP #4/12 87.5G) 0702-static-run-2, 0702-static-run-dark-1, 0702-static-run-ddark-1, 0702-static-walk-1, 0702-static-walk-2, 0702-static-walk-dark-1, 0702-static-walk-ddark-1, 0702-static-walktoward-ddark-1, 0723-calib8mm-1, 0723-calib8mm-2, 0723-calib8mm-3, 0723-calib16mm-1, 0723-calib16mm-2, 0723-calib16mm-3, 0723-calib16mm-4
│ ├── (ZIP #5/12 97.7G) 0815-warf-1, 0815-warf-2, 0815-warf-bright
│ ├── (ZIP #6/12 111.4G) 0815-warf-long, 0815-warf-slow, 0905-ball-mohit-l0, 0905-ball-mohit-l1, 0905-ball-mohit-l2, 0905-ball-sizhuo-l0, 0905-ball-sizhuo-l2, 0905-hdr-sizhuo-f13d2, 0905-hdr-sizhuo-f16
│ ├── (ZIP #7/12 91.1G) 0905-hdr-sizhuo-f16-0, 0905-jump-mohit-l0, 0905-jump-mohit-l1, 0905-jump-mohit-l2, 0905-jump-sizhuo-l0, 0905-jump-sizhuo-l1, 0905-jump-sizhuo-l2, 0905-walk-mohit-l0, 0905-walk-mohit-l1, 0905-walk-mohit-l2, 0905-walk-sizhuo-l0, 0905-walk-sizhuo-l1, 0905-walk-sizhuo-l2, 1005-ocr-far-l1, 1005-ocr-far-l1-test, 1005-ocr-far-l2, 1005-ocr-far-strobe, 1005-ocr-far-strobe-2, 1005-ocr-far-strobe-3, 1005-ocr-near-l1, 1005-ocr-near-l1-2, 1005-ocr-near-l2, 1005-ocr-near-l2-2, 1005-ocr-near-strobe, 1005-ocr-near-strobe-2
│ ├── (ZIP #8/12 29.8G) 1007-bike-1, 1007-bike-2
│ ├── (ZIP #9/12 137.5G) 1007-drive-1
│ ├── (ZIP #10/12 139.0G) 1007-drive-2
│ ├── (ZIP #11/12 108.1G) 1007-walk-1, 1007-walk-2, 1007-walk-3, 1014-slam-l0, 1014-slam-l0-2, 1014-slam-l0-3, 1014-slam-l0-4, 1014-slam-l0-5, 1014-slam-l1
│ └── (ZIP #12/12 49.6G) 1014-slam-l2, 1014-slam-l2-2, 1014-slam-l3, 1014-slam-l4
└── 📄 README.md
Note: The zip file sizes refer to the decompressed filesize.
This dataset was created for an ongoing single photon reconstruction challenge and competition and consists of photoncube/image pairs from 50 unique simulated scenes, plus another 5 scenes for the test set (for which ground truths are not made public). Each photoncube consists of 1024 bitplanes, and the associated ground truth reconstruction corresponds to the last bitplane. A sample of this dataset can be downloaded here (~3.5GB).
To download the whole dataset (~425G training set + ~42G test set, ~133G and ~13G compressed respectively) use DATASET_PREFIX=challenges/reconstruction.
See Detailed Folder Structure
Tree<'train.json'>
├── 💾 train_0.zip (8.5G 14.0x)
├── 💾 train_1.zip (8.5G 4.1x)
├── 💾 train_2.zip (8.5G 3.0x)
├── 💾 train_3.zip (8.5G 4.1x)
├── 💾 train_4.zip (8.5G 2.2x)
├── 💾 train_5.zip (8.5G 22.1x)
├── 💾 train_6.zip (8.5G 4.3x)
├── 💾 train_7.zip (8.5G 5.3x)
├── 💾 train_8.zip (8.5G 2.1x)
├── 💾 train_9.zip (8.5G 2.3x)
├── 💾 train_10.zip (8.5G 2.9x)
├── 💾 train_11.zip (8.5G 6.1x)
├── 💾 train_12.zip (8.5G 3.5x)
├── 💾 train_13.zip (8.5G 3.6x)
├── 💾 train_14.zip (8.5G 3.1x)
├── 💾 train_15.zip (8.5G 2.3x)
├── 💾 train_16.zip (8.5G 2.8x)
├── 💾 train_17.zip (8.5G 5.3x)
├── 💾 train_18.zip (8.5G 2.7x)
├── 💾 train_19.zip (8.5G 2.6x)
├── 💾 train_20.zip (8.5G 3.4x)
├── 💾 train_21.zip (8.5G 4.4x)
├── 💾 train_22.zip (8.5G 2.9x)
├── 💾 train_23.zip (8.5G 3.2x)
├── 💾 train_24.zip (8.5G 5.4x)
├── 💾 train_25.zip (8.5G 3.4x)
├── 💾 train_26.zip (8.5G 2.2x)
├── 💾 train_27.zip (8.5G 4.2x)
├── 💾 train_28.zip (8.5G 2.6x)
├── 💾 train_29.zip (8.5G 2.2x)
├── 💾 train_30.zip (8.5G 3.4x)
├── 💾 train_31.zip (8.5G 2.9x)
├── 💾 train_32.zip (8.5G 2.2x)
├── 💾 train_33.zip (8.5G 2.4x)
├── 💾 train_34.zip (8.5G 3.3x)
├── 💾 train_35.zip (8.5G 2.6x)
├── 💾 train_36.zip (8.5G 1.7x)
├── 💾 train_37.zip (8.5G 3.9x)
├── 💾 train_38.zip (8.5G 3.9x)
├── 💾 train_39.zip (8.5G 3.7x)
├── 💾 train_40.zip (8.5G 2.0x)
├── 💾 train_41.zip (8.5G 5.1x)
├── 💾 train_42.zip (8.5G 2.3x)
├── 💾 train_43.zip (8.5G 5.2x)
├── 💾 train_44.zip (8.5G 7.1x)
├── 💾 train_45.zip (8.5G 2.3x)
├── 💾 train_46.zip (8.5G 3.8x)
├── 💾 train_47.zip (8.5G 2.5x)
├── 💾 train_48.zip (8.5G 3.3x)
╰── 💾 train_49.zip (8.5G 6.5x)
Tree<'test.json'>
├── 💾 test_0.zip (8.5G 8.8x)
├── 💾 test_1.zip (8.5G 1.4x)
├── 💾 test_2.zip (8.5G 9.7x)
├── 💾 test_3.zip (8.5G 2.3x)
╰── 💾 test_4.zip (8.5G 6.9x)
Note: The zip file sizes refer to the decompressed filesize, compression ratio is shown in parenthesis.
Tip: To see the full details for each split, you can use the show-tree like so python tools.py show-tree trees/challenges/reconstruction/test.json --full.
Using the visionsim framework you can simulate large scale datasets with a wide range of ground truth annotations and realistic sensor emulations. Here we provide access to the dataset which was created as part of this tutorial. It contains 50 indoor scenes with realistic camera motion which are animated for 12 seconds and rendered at 100fps at a resolution of 800x800 pixels. Ground truth annotation for metric depths, normals, optical flow (both forward and backwards), object segmentations, as well as camera intrinsics and extrinsics are provided for every frame.
This dataset has been split by types of ground truth annotations before being uploaded, so for instance, you'll find all the depth maps under visionsim/visionsim50/depths, RGB frames under visionsim/visionsim50/frames, etc. The previews folder contains video previews of all scenes and ground truths, and the metadata folder has all the transforms.json files which contain camera intrinsics and extrinsics.
For instance, to download only the RGB data and camera trajectories, you can run the above script with DATASET_PREFIX=visionsim/visionsim50/frames and again with DATASET_PREFIX=visionsim/visionsim50/metadata.
Note: This is a pre-release dataset and is subject to change or get updated.
See Detailed Folder Structure
Tree<'frames.json'>
╰── 📁 datasets (44.5G)
╰── 📁 renders (44.5G)
├── 💾 lazienka_0.zip (775.4M)
├── 💾 game-room_0.zip (846.9M)
├── 💾 tv-couch_0.zip (572.5M)
├── 💾 bathroom4_0.zip (950.7M)
├── 💾 interior-scene_0.zip (912.5M)
├── 💾 loft_0.zip (714.6M)
├── 💾 designer-bedroom_0.zip (1.0G)
├── 💾 livingroom_0.zip (654.6M)
├── 💾 mesa-concept_0.zip (465.0M)
├── 💾 kitchen2_0.zip (661.6M)
├── 💾 cocina-ii_0.zip (749.1M)
├── 💾 junkshop_0.zip (2.1G)
├── 💾 attic_0.zip (768.4M)
├── 💾 diningroom_0.zip (727.4M)
├── 💾 bathroom2_0.zip (452.2M)
├── 💾 paneled-room-revisited_0.zip (743.5M)
├── 💾 restroom_0.zip (952.2M)
├── 💾 bathroom3_0.zip (1.0G)
├── 💾 domestic-office-table_0.zip (1004.2M)
├── 💾 staircase_0.zip (386.5M)
├── 💾 italianflat_0.zip (2.2G)
├── 💾 white-room_0.zip (2.1G)
├── 💾 stone-shower_0.zip (473.4M)
├── 💾 modern-kitchen_0.zip (645.1M)
├── 💾 bathroom5_0.zip (833.3M)
├── 💾 wooden-staircase_0.zip (853.2M)
├── 💾 classroom_0.zip (878.7M)
├── 💾 minimarket_0.zip (948.0M)
├── 💾 kitchenpack_0.zip (597.2M)
├── 💾 bathtime_0.zip (928.3M)
├── 💾 bedroom2_0.zip (840.7M)
├── 💾 gaffer_0.zip (858.0M)
├── 💾 bachelors-quarters_0.zip (885.9M)
├── 💾 morning-apartment_0.zip (630.6M)
├── 💾 restaurant_0.zip (2.2G)
├── 💾 diner_0.zip (691.7M)
├── 💾 country-kitchen_0.zip (2.1G)
├── 💾 kitchen3_0.zip (416.6M)
├── 💾 barbershop_0.zip (930.2M)
├── 💾 cozykitchen_0.zip (984.4M)
├── 💾 bathroom1_0.zip (779.4M)
├── 💾 library-homeoffice_0.zip (909.6M)
├── 💾 ultramodern_0.zip (827.1M)
├── 💾 kitchen1_0.zip (805.5M)
├── 💾 simplekitchen_0.zip (611.6M)
├── 💾 lynxsdesign_0.zip (860.1M)
├── 💾 bath_0.zip (659.6M)
├── 💾 bedroom1_0.zip (638.5M)
├── 💾 officebuilding_0.zip (872.4M)
╰── 💾 sunny-room_0.zip (761.4M)
Tree<'depths.json'>
╰── 📁 datasets (115.4G)
╰── 📁 renders (115.4G)
├── 💾 lazienka_0.zip (1.7G)
├── 💾 game-room_0.zip (2.0G)
├── 💾 tv-couch_0.zip (2.2G)
├── 💾 bathroom4_0.zip (2.1G)
├── 💾 interior-scene_0.zip (2.2G)
├── 💾 loft_0.zip (2.0G)
├── 💾 designer-bedroom_0.zip (2.0G)
├── 💾 livingroom_0.zip (2.0G)
├── 💾 mesa-concept_0.zip (1.8G)
├── 💾 kitchen2_0.zip (2.0G)
├── 💾 cocina-ii_0.zip (2.1G)
├── 💾 junkshop_0.zip (3.9G)
├── 💾 attic_0.zip (1.9G)
├── 💾 diningroom_0.zip (1.6G)
├── 💾 bathroom2_0.zip (2.4G)
├── 💾 paneled-room-revisited_0.zip (2.0G)
├── 💾 restroom_0.zip (2.1G)
├── 💾 bathroom3_0.zip (2.3G)
├── 💾 domestic-office-table_0.zip (1.6G)
├── 💾 staircase_0.zip (2.2G)
├── 💾 italianflat_0.zip (4.2G)
├── 💾 white-room_0.zip (4.0G)
├── 💾 stone-shower_0.zip (2.2G)
├── 💾 modern-kitchen_0.zip (2.0G)
├── 💾 bathroom5_0.zip (2.0G)
├── 💾 wooden-staircase_0.zip (2.2G)
├── 💾 classroom_0.zip (2.1G)
├── 💾 minimarket_0.zip (2.3G)
├── 💾 kitchenpack_0.zip (2.1G)
├── 💾 bathtime_0.zip (1.9G)
├── 💾 bedroom2_0.zip (2.3G)
├── 💾 gaffer_0.zip (2.0G)
├── 💾 bachelors-quarters_0.zip (2.2G)
├── 💾 morning-apartment_0.zip (2.3G)
├── 💾 restaurant_0.zip (3.9G)
├── 💾 diner_0.zip (2.1G)
├── 💾 country-kitchen_0.zip (5.1G)
├── 💾 kitchen3_0.zip (2.0G)
├── 💾 barbershop_0.zip (2.5G)
├── 💾 cozykitchen_0.zip (2.5G)
├── 💾 bathroom1_0.zip (3.0G)
├── 💾 library-homeoffice_0.zip (2.1G)
├── 💾 ultramodern_0.zip (2.1G)
├── 💾 kitchen1_0.zip (2.2G)
├── 💾 simplekitchen_0.zip (1.8G)
├── 💾 lynxsdesign_0.zip (1.8G)
├── 💾 bath_0.zip (2.0G)
├── 💾 bedroom1_0.zip (2.2G)
├── 💾 officebuilding_0.zip (1.9G)
╰── 💾 sunny-room_0.zip (2.3G)
Tree<'normals.json'>
╰── 📁 datasets (296.2G)
╰── 📁 renders (296.2G)
├── 💾 lazienka_0.zip (1.0G)
├── 💾 game-room_0.zip (3.5G)
├── 💾 tv-couch_0.zip (4.5G)
├── 💾 bathroom4_0.zip (8.6G)
├── 💾 interior-scene_0.zip (7.2G)
├── 💾 loft_0.zip (2.4G)
├── 💾 designer-bedroom_0.zip (8.1G)
├── 💾 livingroom_0.zip (3.1G)
├── 💾 mesa-concept_0.zip (2.0G)
├── 💾 kitchen2_0.zip (2.9G)
├── 💾 cocina-ii_0.zip (6.7G)
├── 💾 junkshop_0.zip (9.4G)
├── 💾 attic_0.zip (5.4G)
├── 💾 diningroom_0.zip (4.0G)
├── 💾 bathroom2_0.zip (2.0G)
├── 💾 paneled-room-revisited_0.zip (4.9G)
├── 💾 restroom_0.zip (8.4G)
├── 💾 bathroom3_0.zip (8.5G)
├── 💾 domestic-office-table_0.zip (8.6G)
├── 💾 staircase_0.zip (1.6G)
├── 💾 italianflat_0.zip (7.9G)
├── 💾 white-room_0.zip (4.4G)
├── 💾 stone-shower_0.zip (3.6G)
├── 💾 modern-kitchen_0.zip (3.3G)
├── 💾 bathroom5_0.zip (4.3G)
├── 💾 wooden-staircase_0.zip (6.9G)
├── 💾 classroom_0.zip (9.4G)
├── 💾 minimarket_0.zip (5.4G)
├── 💾 kitchenpack_0.zip (573.4M)
├── 💾 bathtime_0.zip (7.0G)
├── 💾 bedroom2_0.zip (7.1G)
├── 💾 gaffer_0.zip (7.8G)
├── 💾 bachelors-quarters_0.zip (5.5G)
├── 💾 morning-apartment_0.zip (2.4G)
├── 💾 restaurant_0.zip (10.0G)
├── 💾 diner_0.zip (6.1G)
├── 💾 country-kitchen_0.zip (10.7G)
├── 💾 kitchen3_0.zip (3.0G)
├── 💾 barbershop_0.zip (8.2G)
├── 💾 cozykitchen_0.zip (8.8G)
├── 💾 bathroom1_0.zip (9.0G)
├── 💾 library-homeoffice_0.zip (7.9G)
├── 💾 ultramodern_0.zip (7.6G)
├── 💾 kitchen1_0.zip (7.3G)
├── 💾 simplekitchen_0.zip (4.0G)
├── 💾 lynxsdesign_0.zip (7.8G)
├── 💾 bath_0.zip (7.7G)
├── 💾 bedroom1_0.zip (8.0G)
├── 💾 officebuilding_0.zip (6.1G)
╰── 💾 sunny-room_0.zip (5.5G)
Tree<'flows.json'>
╰── 📁 datasets (459.4G)
╰── 📁 renders (459.4G)
├── 💾 lazienka_0.zip (9.3G)
├── 💾 game-room_0.zip (7.8G)
├── 💾 tv-couch_0.zip (10.2G)
├── 💾 bathroom4_0.zip (8.1G)
├── 💾 interior-scene_0.zip (8.4G)
├── 💾 loft_0.zip (9.7G)
├── 💾 designer-bedroom_0.zip (8.4G)
├── 💾 livingroom_0.zip (10.3G)
├── 💾 mesa-concept_0.zip (9.6G)
├── 💾 kitchen2_0.zip (9.6G)
├── 💾 cocina-ii_0.zip (9.8G)
├── 💾 junkshop_0.zip (9.1G)
├── 💾 attic_0.zip (9.7G)
├── 💾 diningroom_0.zip (9.6G)
├── 💾 bathroom2_0.zip (9.9G)
├── 💾 paneled-room-revisited_0.zip (10.2G)
├── 💾 restroom_0.zip (7.7G)
├── 💾 bathroom3_0.zip (8.1G)
├── 💾 domestic-office-table_0.zip (9.9G)
├── 💾 staircase_0.zip (10.2G)
├── 💾 italianflat_0.zip (9.7G)
├── 💾 white-room_0.zip (11.9G)
├── 💾 stone-shower_0.zip (10.4G)
├── 💾 modern-kitchen_0.zip (10.2G)
├── 💾 bathroom5_0.zip (7.6G)
├── 💾 wooden-staircase_0.zip (9.6G)
├── 💾 classroom_0.zip (4.4G)
├── 💾 minimarket_0.zip (10.3G)
├── 💾 kitchenpack_0.zip (5.5G)
├── 💾 bathtime_0.zip (9.9G)
├── 💾 bedroom2_0.zip (8.3G)
├── 💾 gaffer_0.zip (9.1G)
├── 💾 bachelors-quarters_0.zip (10.2G)
├── 💾 morning-apartment_0.zip (9.9G)
├── 💾 restaurant_0.zip (9.8G)
├── 💾 diner_0.zip (10.1G)
├── 💾 country-kitchen_0.zip (13.7G)
├── 💾 kitchen3_0.zip (7.2G)
├── 💾 barbershop_0.zip (10.0G)
├── 💾 cozykitchen_0.zip (8.4G)
├── 💾 bathroom1_0.zip (11.7G)
├── 💾 library-homeoffice_0.zip (7.6G)
├── 💾 ultramodern_0.zip (10.7G)
├── 💾 kitchen1_0.zip (8.6G)
├── 💾 simplekitchen_0.zip (7.9G)
├── 💾 lynxsdesign_0.zip (8.6G)
├── 💾 bath_0.zip (10.0G)
├── 💾 bedroom1_0.zip (7.6G)
├── 💾 officebuilding_0.zip (10.0G)
╰── 💾 sunny-room_0.zip (5.2G)
Tree<'segmentations.json'>
╰── 📁 datasets (27.4G)
╰── 📁 renders (27.4G)
├── 💾 lazienka_0.zip (58.6M)
├── 💾 game-room_0.zip (702.8M)
├── 💾 tv-couch_0.zip (49.1M)
├── 💾 bathroom4_0.zip (712.1M)
├── 💾 interior-scene_0.zip (176.5M)
├── 💾 loft_0.zip (47.8M)
├── 💾 designer-bedroom_0.zip (655.4M)
├── 💾 livingroom_0.zip (68.4M)
├── 💾 mesa-concept_0.zip (49.9M)
├── 💾 kitchen2_0.zip (80.4M)
├── 💾 cocina-ii_0.zip (61.4M)
├── 💾 junkshop_0.zip (2.3G)
├── 💾 attic_0.zip (84.7M)
├── 💾 diningroom_0.zip (75.2M)
├── 💾 bathroom2_0.zip (38.7M)
├── 💾 paneled-room-revisited_0.zip (96.1M)
├── 💾 restroom_0.zip (113.8M)
├── 💾 bathroom3_0.zip (608.6M)
├── 💾 domestic-office-table_0.zip (90.7M)
├── 💾 staircase_0.zip (56.6M)
├── 💾 italianflat_0.zip (2.3G)
├── 💾 white-room_0.zip (2.2G)
├── 💾 stone-shower_0.zip (65.2M)
├── 💾 modern-kitchen_0.zip (72.4M)
├── 💾 bathroom5_0.zip (644.7M)
├── 💾 wooden-staircase_0.zip (100.1M)
├── 💾 classroom_0.zip (161.8M)
├── 💾 minimarket_0.zip (89.7M)
├── 💾 kitchenpack_0.zip (187.9M)
├── 💾 bathtime_0.zip (522.7M)
├── 💾 bedroom2_0.zip (720.1M)
├── 💾 gaffer_0.zip (160.4M)
├── 💾 bachelors-quarters_0.zip (88.4M)
├── 💾 morning-apartment_0.zip (108.8M)
├── 💾 restaurant_0.zip (2.3G)
├── 💾 diner_0.zip (90.1M)
├── 💾 country-kitchen_0.zip (5.0G)
├── 💾 kitchen3_0.zip (81.0M)
├── 💾 barbershop_0.zip (168.1M)
├── 💾 cozykitchen_0.zip (833.0M)
├── 💾 bathroom1_0.zip (2.9G)
├── 💾 library-homeoffice_0.zip (143.2M)
├── 💾 ultramodern_0.zip (592.7M)
├── 💾 kitchen1_0.zip (703.7M)
├── 💾 simplekitchen_0.zip (120.9M)
├── 💾 lynxsdesign_0.zip (695.1M)
├── 💾 bath_0.zip (67.6M)
├── 💾 bedroom1_0.zip (127.0M)
├── 💾 officebuilding_0.zip (119.4M)
╰── 💾 sunny-room_0.zip (149.2M)
Tree<'previews.json'>
╰── 💾 datasets_0.zip (6.5G)
Tree<'metadata.json'>
╰── 💾 datasets_0.zip (51.2M)
Note: The zip file sizes refer to the decompressed filesize.
Tip: To see the full details for each split, you can use the show-tree like so python tools.py show-tree trees/visionsim50/frames.json --full.
We include high-speed video sequences (at framerates of about 16 kFPS) and groundtruth annotations for three computer vision tasks: monocular depth estimation, multi-frame point tracking, and video restoration (for which the input high-speed sequences are the groundtruth itself). These high-speed sequences were used to simulate photon detections and train Quanta Neural Networks, published at ICCV 2025.
To download these datasets, please use DATASET_PREFIX=quanta-neural-networks.
See Detailed Folder Structure
Tree<'tree.json'>
╰── 💾 blender_depth_faster_0.zip (22.3G 1.1x)
Tree<'tree.json'>
╰── 💾 tracking_dataset_0.zip (8.7G 1.8x)
Tree<'tree.json'>
╰── 💾 xvfi_0.zip (1.7G 1.0x)
Note: The zip file sizes refer to the decompressed filesize.
Tip: To see the full details for each split, you can use the show-tree like so python tools.py show-tree trees/quanta-neural-networks/xvfi.json --full.