Dataset Viewer
Auto-converted to Parquet Duplicate
traj_idx
int32
0
652
num_steps
int32
16
1.1k
trajectory
listlengths
16
1.08k
task_instruction
stringclasses
35 values
motion_description
stringlengths
11
141
video_raw
video
video_trajviz
video
last_frame_raw
imagewidth (px)
200
640
last_frame_trajviz
imagewidth (px)
200
640
opticalflow
imagewidth (px)
208
640
storyboard_key2
imagewidth (px)
508
508
storyboard_key2_trajviz
imagewidth (px)
508
508
storyboard_key4
imagewidth (px)
558
558
storyboard_key4_trajviz
imagewidth (px)
558
558
storyboard_key9
imagewidth (px)
822
822
storyboard_key9_trajviz
imagewidth (px)
822
822
storyboard_key16
imagewidth (px)
1.09k
1.09k
storyboard_key16_trajviz
imagewidth (px)
1.09k
1.09k
0
682
[ [ 90, 98 ], [ 91, 98 ], [ 91, 98 ], [ 91, 98 ], [ 91, 98 ], [ 99, 69 ], [ 104, 65 ], [ 110, 63 ], [ 116, 62 ], [ 122, 61 ], [ 127, 62 ], [ 131, 62 ], [ 135, 61 ],...
brush hair
move downward and move upward, repeating this sequence 3 times
Not supported with pagination yet
1
256
[ [ 74, 310 ], [ 75, 318 ], [ 79, 316 ], [ 83, 314 ], [ 87, 311 ], [ 91, 309 ], [ 95, 306 ], [ 100, 304 ], [ 103, 302 ], [ 107, 299 ], [ 110, 297 ], [ 113, 295 ], [ 115, ...
remove hair tie
move to the top right and move to the bottom left
2
316
[ [ 52, 268 ], [ 57, 265 ], [ 62, 261 ], [ 66, 257 ], [ 71, 253 ], [ 75, 249 ], [ 80, 244 ], [ 83, 239 ], [ 87, 233 ], [ 91, 228 ], [ 95, 222 ], [ 99, 216 ], [ 103, 2...
remove hair tie
move upward and to the right, move downward, and move upward and to the left
3
202
[ [ 74, 311 ], [ 84, 304 ], [ 92, 298 ], [ 101, 293 ], [ 109, 289 ], [ 115, 284 ], [ 121, 279 ], [ 129, 271 ], [ 134, 262 ], [ 141, 254 ], [ 147, 244 ], [ 155, 232 ], [ 1...
remove hair tie
move upward and to the right, then move downward and to the left
4
473
[ [ 33, 260 ], [ 36, 245 ], [ 39, 236 ], [ 43, 228 ], [ 45, 221 ], [ 46, 217 ], [ 45, 214 ], [ 44, 211 ], [ 37, 210 ], [ 32, 207 ], [ 27, 207 ], [ 24, 207 ], [ 23, 20...
erase whiteboard
move to the right and move to the left, repeating this sequence 2 times
Not supported with pagination yet
5
527
[ [ 41, 283 ], [ 41, 283 ], [ 51, 252 ], [ 51, 252 ], [ 57, 229 ], [ 52, 217 ], [ 50, 207 ], [ 49, 200 ], [ 49, 191 ], [ 53, 183 ], [ 58, 174 ], [ 63, 165 ], [ 67, 15...
erase whiteboard
move upward and downward, repeating this sequence for 5 times, and then move to the bottom left
Not supported with pagination yet
6
481
[ [ 17, 275 ], [ 16, 279 ], [ 16, 285 ], [ 16, 288 ], [ 17, 293 ], [ 15, 296 ], [ 15, 296 ], [ 15, 296 ], [ 15, 296 ], [ 15, 296 ], [ 15, 296 ], [ 15, 296 ], [ 15, 29...
erase whiteboard
move upward and move to the right, then move to the left while making vertical oscillations
Not supported with pagination yet
7
74
[ [ 391, 350 ], [ 384, 348 ], [ 382, 350 ], [ 381, 351 ], [ 380, 351 ], [ 380, 352 ], [ 379, 353 ], [ 379, 353 ], [ 378, 354 ], [ 378, 353 ], [ 378, 353 ], [ 378, 354 ], [ ...
handover knife
move to the left
8
80
[ [ 25, 393 ], [ 31, 391 ], [ 37, 389 ], [ 42, 388 ], [ 48, 386 ], [ 54, 385 ], [ 60, 384 ], [ 63, 384 ], [ 68, 383 ], [ 72, 383 ], [ 77, 383 ], [ 81, 383 ], [ 86, 38...
handover knife
move to the right
9
172
[ [ 31, 472 ], [ 45, 464 ], [ 46, 463 ], [ 46, 463 ], [ 46, 463 ], [ 44, 460 ], [ 44, 459 ], [ 43, 459 ], [ 42, 459 ], [ 41, 458 ], [ 39, 457 ], [ 40, 458 ], [ 38, 45...
handover knife
move upward and to the right, then shortly move to the left
10
110
[ [ 415, 420 ], [ 413, 419 ], [ 414, 419 ], [ 413, 419 ], [ 413, 419 ], [ 413, 419 ], [ 412, 419 ], [ 413, 420 ], [ 413, 420 ], [ 413, 421 ], [ 412, 420 ], [ 412, 421 ], [ ...
handover knife
move to the left and upward, then rotate
11
80
[ [ 369, 375 ], [ 369, 375 ], [ 368, 375 ], [ 368, 375 ], [ 368, 375 ], [ 370, 373 ], [ 370, 373 ], [ 370, 373 ], [ 370, 373 ], [ 370, 373 ], [ 370, 373 ], [ 370, 373 ], [ ...
handover plate
move to the left
12
70
[ [ 92, 372 ], [ 93, 371 ], [ 97, 370 ], [ 98, 371 ], [ 99, 371 ], [ 104, 369 ], [ 103, 370 ], [ 107, 371 ], [ 110, 371 ], [ 113, 370 ], [ 117, 372 ], [ 123, 371 ], [ 127...
handover plate
move to the right
13
80
[ [ 187, 382 ], [ 187, 382 ], [ 187, 382 ], [ 187, 382 ], [ 187, 382 ], [ 187, 382 ], [ 187, 382 ], [ 212, 394 ], [ 215, 393 ], [ 219, 391 ], [ 222, 389 ], [ 222, 391 ], [ ...
handover plate
move downward and to the right
14
105
[ [ 387, 159 ], [ 389, 160 ], [ 389, 160 ], [ 389, 161 ], [ 389, 160 ], [ 390, 160 ], [ 391, 160 ], [ 391, 160 ], [ 391, 160 ], [ 392, 159 ], [ 392, 159 ], [ 392, 159 ], [ ...
handover plate
move downward and to the left
15
111
[ [ 129, 437 ], [ 130, 430 ], [ 134, 425 ], [ 135, 422 ], [ 140, 416 ], [ 140, 411 ], [ 145, 405 ], [ 147, 401 ], [ 148, 394 ], [ 151, 390 ], [ 150, 387 ], [ 149, 383 ], [ ...
handover plate
move upward to the right
16
160
[ [ 456, 115 ], [ 452, 111 ], [ 447, 109 ], [ 444, 108 ], [ 442, 105 ], [ 439, 104 ], [ 437, 102 ], [ 436, 100 ], [ 434, 99 ], [ 431, 98 ], [ 430, 99 ], [ 429, 97 ], [ 42...
pick up the cup and place it to the lower left of the laptop
move to the bottom left
17
188
[ [ 461, 139 ], [ 459, 134 ], [ 458, 131 ], [ 455, 128 ], [ 452, 125 ], [ 450, 122 ], [ 447, 119 ], [ 444, 118 ], [ 440, 116 ], [ 438, 114 ], [ 435, 112 ], [ 433, 109 ], [ ...
pick up the cup and place it to the lower left of the laptop
move downward and then move to the left
18
123
[ [ 452, 139 ], [ 446, 132 ], [ 439, 126 ], [ 434, 124 ], [ 430, 119 ], [ 427, 116 ], [ 424, 112 ], [ 421, 110 ], [ 419, 107 ], [ 418, 107 ], [ 417, 107 ], [ 415, 104 ], [ ...
pick up the cup and place it to the lower left of the laptop
move downward, getting closer to the laptop, and then move to the left
19
90
[ [ 417, 109 ], [ 420, 115 ], [ 420, 115 ], [ 420, 115 ], [ 419, 114 ], [ 419, 116 ], [ 418, 114 ], [ 419, 113 ], [ 418, 112 ], [ 420, 112 ], [ 421, 113 ], [ 422, 112 ], [ ...
pick up the cup and place it to the lower left of the laptop
move downward and to the left
20
102
[ [ 429, 111 ], [ 425, 112 ], [ 430, 115 ], [ 429, 113 ], [ 429, 112 ], [ 429, 114 ], [ 429, 116 ], [ 431, 114 ], [ 434, 116 ], [ 433, 118 ], [ 434, 120 ], [ 434, 124 ], [ ...
pick up the cup and place it to the lower left of the laptop
move downward, getting closer to the laptop, and then move to the left
21
137
[ [ 588, 225 ], [ 588, 238 ], [ 588, 237 ], [ 587, 237 ], [ 587, 236 ], [ 586, 234 ], [ 584, 230 ], [ 581, 224 ], [ 578, 216 ], [ 576, 206 ], [ 574, 195 ], [ 571, 183 ], [ ...
pick up the cup and place it to the lower left of the laptop
move upward shortly, move downward, and then move to the left
22
229
[ [ 218, 157 ], [ 219, 160 ], [ 217, 162 ], [ 218, 162 ], [ 217, 164 ], [ 217, 165 ], [ 216, 166 ], [ 215, 167 ], [ 215, 168 ], [ 215, 168 ], [ 215, 168 ], [ 214, 168 ], [ ...
stir
make 3 circular motions clockwise
23
135
[ [ 216, 172 ], [ 224, 170 ], [ 226, 168 ], [ 227, 168 ], [ 229, 168 ], [ 230, 170 ], [ 229, 170 ], [ 229, 170 ], [ 232, 169 ], [ 233, 167 ], [ 232, 163 ], [ 233, 159 ], [ ...
stir
move to the left shortly, and vertically move down and up
24
125
[ [ 198, 173 ], [ 199, 164 ], [ 198, 162 ], [ 201, 160 ], [ 204, 157 ], [ 204, 157 ], [ 215, 147 ], [ 220, 142 ], [ 226, 137 ], [ 229, 133 ], [ 233, 130 ], [ 236, 128 ], [ ...
stir
make a half circular motion counterclockwise, then reverse it by moving clockwise
25
100
[ [ 182, 207 ], [ 209, 152 ], [ 210, 150 ], [ 212, 150 ], [ 212, 150 ], [ 212, 155 ], [ 208, 156 ], [ 205, 162 ], [ 204, 168 ], [ 198, 174 ], [ 195, 183 ], [ 191, 188 ], [ ...
stir
move upward and downward, repeating this sequence 2.5 times
26
349
[ [ 185, 206 ], [ 191, 204 ], [ 187, 206 ], [ 181, 211 ], [ 177, 214 ], [ 174, 217 ], [ 172, 220 ], [ 170, 223 ], [ 169, 226 ], [ 169, 227 ], [ 169, 229 ], [ 171, 231 ], [ ...
stir
move downward to the right and reverse, move upward to the right and reverse, repeating this sequence 3 times
27
335
[ [ 153, 221 ], [ 151, 222 ], [ 150, 222 ], [ 150, 222 ], [ 150, 223 ], [ 150, 223 ], [ 149, 222 ], [ 149, 222 ], [ 150, 222 ], [ 150, 222 ], [ 150, 221 ], [ 151, 220 ], [ ...
stir
move closer to the top right part of the bowl and move farther from that point, repeating this sequence 5 times
28
70
[ [ 165, 179 ], [ 182, 173 ], [ 181, 185 ], [ 178, 193 ], [ 175, 202 ], [ 178, 211 ], [ 166, 221 ], [ 166, 221 ], [ 166, 221 ], [ 150, 241 ], [ 146, 246 ], [ 144, 246 ], [ ...
stir
move downward, upward, and downward
29
80
[ [ 102, 60 ], [ 110, 45 ], [ 109, 33 ], [ 106, 32 ], [ 104, 25 ], [ 101, 25 ], [ 97, 27 ], [ 93, 29 ], [ 88, 35 ], [ 87, 36 ], [ 83, 42 ], [ 82, 45 ], [ 81, 49 ], ...
stir
make a circular motion counter-clockwise, then move to the right
30
274
[ [ 83, 60 ], [ 86, 57 ], [ 90, 58 ], [ 91, 54 ], [ 89, 53 ], [ 86, 53 ], [ 83, 55 ], [ 81, 56 ], [ 78, 55 ], [ 76, 56 ], [ 73, 57 ], [ 70, 57 ], [ 68, 55 ], [ ...
stir
make a circular motion counter-clockwise, then move to the right
31
90
[ [ 110, 85 ], [ 116, 82 ], [ 122, 80 ], [ 122, 82 ], [ 118, 84 ], [ 116, 87 ], [ 112, 92 ], [ 110, 98 ], [ 106, 102 ], [ 101, 109 ], [ 96, 117 ], [ 98, 124 ], [ 96, ...
stir
move downward and upward, repeating this sequence 2.5 times
32
48
[ [ 236, 152 ], [ 224, 168 ], [ 227, 170 ], [ 229, 173 ], [ 230, 177 ], [ 231, 180 ], [ 232, 183 ], [ 233, 187 ], [ 234, 190 ], [ 235, 192 ], [ 233, 195 ], [ 233, 196 ], [ ...
stir
move downward
33
175
[ [ 308, 201 ], [ 294, 202 ], [ 281, 202 ], [ 272, 201 ], [ 262, 201 ], [ 255, 200 ], [ 249, 200 ], [ 243, 201 ], [ 238, 200 ], [ 233, 201 ], [ 228, 203 ], [ 224, 203 ], [ ...
stir
make 2 circular motions counter-clockwise
34
89
[ [ 179, 203 ], [ 186, 231 ], [ 190, 235 ], [ 192, 240 ], [ 195, 244 ], [ 199, 249 ], [ 201, 252 ], [ 204, 256 ], [ 207, 259 ], [ 209, 262 ], [ 211, 265 ], [ 213, 268 ], [ ...
stir
move downward, move upward, then move downward and to the right
35
35
[ [ 258, 251 ], [ 258, 251 ], [ 258, 251 ], [ 258, 251 ], [ 258, 251 ], [ 258, 251 ], [ 258, 251 ], [ 258, 251 ], [ 165, 215 ], [ 169, 215 ], [ 173, 218 ], [ 177, 219 ], [ ...
stir
move upward and to the left, then move downward and to the right
36
118
[ [ 417, 157 ], [ 413, 151 ], [ 409, 146 ], [ 406, 142 ], [ 403, 138 ], [ 401, 135 ], [ 399, 129 ], [ 398, 127 ], [ 397, 126 ], [ 395, 124 ], [ 395, 124 ], [ 394, 125 ], [ ...
pick up the cup and place it to the lower left of the laptop
move downward and to the left
37
216
[ [ 263, 183 ], [ 263, 183 ], [ 288, 225 ], [ 288, 225 ], [ 288, 225 ], [ 288, 225 ], [ 288, 225 ], [ 288, 225 ], [ 288, 225 ], [ 288, 225 ], [ 288, 225 ], [ 288, 225 ], [ ...
pick up the cup and place it to the lower left of the laptop
move to the left
38
216
[ [ 409, 140 ], [ 409, 140 ], [ 409, 140 ], [ 409, 140 ], [ 409, 140 ], [ 409, 140 ], [ 409, 140 ], [ 409, 140 ], [ 409, 140 ], [ 409, 140 ], [ 409, 140 ], [ 409, 140 ], [ ...
pick up the cup and place it to the lower left of the laptop
move downward, then move to the left following a convex curve
39
139
[ [ 261, 176 ], [ 260, 177 ], [ 261, 176 ], [ 259, 176 ], [ 261, 175 ], [ 261, 176 ], [ 261, 175 ], [ 262, 176 ], [ 260, 175 ], [ 263, 174 ], [ 262, 175 ], [ 262, 174 ], [ ...
pick up the cup and place it to the lower left of the laptop
move upward, then move downward to the left
40
137
[ [ 282, 188 ], [ 282, 188 ], [ 279, 176 ], [ 279, 178 ], [ 272, 181 ], [ 271, 182 ], [ 269, 182 ], [ 267, 183 ], [ 263, 184 ], [ 263, 184 ], [ 263, 184 ], [ 267, 186 ], [ ...
pick up the cup and place it to the lower left of the laptop
move to the left
41
137
[ [ 423, 169 ], [ 420, 162 ], [ 415, 159 ], [ 410, 155 ], [ 408, 153 ], [ 408, 152 ], [ 406, 152 ], [ 405, 151 ], [ 405, 150 ], [ 404, 149 ], [ 403, 149 ], [ 403, 149 ], [ ...
pick up the cup and place it to the lower left of the laptop
move downward, then move to the left
42
121
[ [ 163, 147 ], [ 177, 141 ], [ 177, 141 ], [ 177, 141 ], [ 177, 141 ], [ 177, 141 ], [ 177, 141 ], [ 177, 141 ], [ 177, 141 ], [ 179, 178 ], [ 178, 181 ], [ 177, 184 ], [ ...
brush hair
move downward
End of preview. Expand in Data Studio

MotIF-1K Dataset

Multimodal trajectories of human and Stretch-robot motion paired with task and motion annotations, released with the paper "MotIF: Motion Instruction Fine-tuning".

Abstract

Many robotics tasks require observing the full motion of the robot — not just start/end states — to correctly judge success (e.g., brushing hair requires the right trajectory, not just ending up "at" the hair). Off-the-shelf vision-language models (VLMs) struggle here because they are trained on single frames and lack robot-motion data. MotIF fine-tunes VLMs using abstract visual motion representations (e.g., keypoint trajectories overlaid on the initial frame, optical flow, key-frame storyboards) to semantically ground robot behavior in the environment. On MotIF-1K, the resulting model outperforms state-of-the-art VLMs by ≥2× in precision and by 56.1% in recall, generalizing across unseen motions, tasks, and environments.

Configs

Config Rows Source
human_motion 653 Human demonstrations (brushing, pouring, etc.)
stretch_motion 370 Hello Robot Stretch teleop demonstrations

Covers 13 task categories with varied feasible motions per task.

Fields

  • traj_idx (int) — trajectory index
  • num_steps (int) — number of frames
  • trajectory (list[list[int]]) — 2D keypoint path [x, y] per frame
  • task_instruction (str) — high-level task (e.g. "brush hair")
  • motion_description (str) — motion verb phrase (e.g. "move downward and upward, repeating 3 times")
  • video_raw, video_trajviz (video) — original and trajectory-overlaid clips (mp4)
  • last_frame_raw, last_frame_trajviz (image) — final-frame stills
  • opticalflow (image) — optical-flow visualization (some rows may be null in human_motion)
  • storyboard_key{2,4,9,16} (image) — storyboard grids at K key-frames
  • storyboard_key{2,4,9,16}_trajviz (image) — same grids with trajectory overlay

Usage

from datasets import load_dataset

ds = load_dataset("myconnects/motif", "human_motion", split="train")
row = ds[0]
print(row["task_instruction"], row["motion_description"])
row["video_raw"]              # decoded video
row["last_frame_trajviz"]     # PIL Image

Training & Evaluation Code

The MotIF GitHub repository contains the full codebase: data-collection scripts, LoRA fine-tuning pipeline (LLaVA-based), evaluation (with optional logits), pretrained model checkpoints, a videoLM-architecture variant, a success-detection comparison (GPT / Gemini / MotIF), and a Gradio web UI.

Citation

@article{hwang2024motif,
  title     = {MotIF: Motion Instruction Fine-tuning},
  author    = {Hwang, Minyoung and Hejna, Joey and Sadigh, Dorsa and Bisk, Yonatan},
  booktitle = {arXiv preprint arXiv:2409.10683},
  year      = {2024},
}

Acknowledgements

We thank Abitha Thankaraj, Hao Zhu, Leena Mathur, Quanting Xie, Rosa Vitiello, Su Li, Tiffany Min, Vidhi Jain, and Yingshan Chang for helping us collect the dataset and providing thoughtful feedback.

Downloads last month
13

Paper for myconnects/motif