Dataset Viewer
id
int64 1.34k
3.18k
| dataset
stringclasses 3
values | scene_name
stringlengths 8
12
| question_type
stringclasses 1
value | category
stringlengths 2
12
| question
stringlengths 831
859
| options
sequencelengths 4
4
| ground_truth
stringlengths 3
12
| mc_answer
stringclasses 4
values | video_suffix
stringclasses 2
values | version
stringclasses 1
value | category, first object
null | category, second object
null | category, positioning object
null | category, orienting object
null | category, querying object
null |
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
1,338 |
arkitscenes
|
42446103
|
object_rel_distance
|
stool
|
[Task]
This video captures an indoor scene. Your task is to first understand the spatial layout of the scene and then represent it using a 10x10 matrix.
[Rule]
1. We provide the cared categories and their corresponding instances: 1 stool, 3 chair, 1 stove, 2 table, 1 tv in this scene. Focus ONLY on these categories.
2. You should use 0 to represent an empty grid and different numbers to represent different instances.
3. If a category contains multiple instances, include all of them in the matrix.
4. If objects overlap in the same grid, retain only the object with the largest area.
[Output]
Fill in the locations of ALL objects (at the instance level) in the matrix.
STRICTLY follow this exact JSON format for your response:
{"grid": 10x10 matrix (list of lists), "legend": {0: "empty", 1: "object 1", 2: "object 2", ...}}
|
[
"A. chair",
"B. stove",
"C. table",
"D. tv"
] |
chair
|
A
|
.mov
|
241009
| null | null | null | null | null |
1,358 |
arkitscenes
|
42897629
|
object_rel_distance
|
stove
|
[Task]
This video captures an indoor scene. Your task is to first understand the spatial layout of the scene and then represent it using a 10x10 matrix.
[Rule]
1. We provide the cared categories and their corresponding instances: 1 stove, 2 washer, 1 tv, 2 sofa, 4 table in this scene. Focus ONLY on these categories.
2. You should use 0 to represent an empty grid and different numbers to represent different instances.
3. If a category contains multiple instances, include all of them in the matrix.
4. If objects overlap in the same grid, retain only the object with the largest area.
[Output]
Fill in the locations of ALL objects (at the instance level) in the matrix.
STRICTLY follow this exact JSON format for your response:
{"grid": 10x10 matrix (list of lists), "legend": {0: "empty", 1: "object 1", 2: "object 2", ...}}
|
[
"A. washer",
"B. tv",
"C. sofa",
"D. table"
] |
table
|
D
|
.mov
|
241009
| null | null | null | null | null |
1,359 |
arkitscenes
|
42897629
|
object_rel_distance
|
tv
|
[Task]
This video captures an indoor scene. Your task is to first understand the spatial layout of the scene and then represent it using a 10x10 matrix.
[Rule]
1. We provide the cared categories and their corresponding instances: 1 tv, 4 table, 1 refrigerator, 2 washer, 1 stove in this scene. Focus ONLY on these categories.
2. You should use 0 to represent an empty grid and different numbers to represent different instances.
3. If a category contains multiple instances, include all of them in the matrix.
4. If objects overlap in the same grid, retain only the object with the largest area.
[Output]
Fill in the locations of ALL objects (at the instance level) in the matrix.
STRICTLY follow this exact JSON format for your response:
{"grid": 10x10 matrix (list of lists), "legend": {0: "empty", 1: "object 1", 2: "object 2", ...}}
|
[
"A. table",
"B. refrigerator",
"C. washer",
"D. stove"
] |
table
|
A
|
.mov
|
241009
| null | null | null | null | null |
1,373 |
arkitscenes
|
42898849
|
object_rel_distance
|
stove
|
[Task]
This video captures an indoor scene. Your task is to first understand the spatial layout of the scene and then represent it using a 10x10 matrix.
[Rule]
1. We provide the cared categories and their corresponding instances: 1 stove, 1 tv, 1 fireplace, 4 table, 8 chair in this scene. Focus ONLY on these categories.
2. You should use 0 to represent an empty grid and different numbers to represent different instances.
3. If a category contains multiple instances, include all of them in the matrix.
4. If objects overlap in the same grid, retain only the object with the largest area.
[Output]
Fill in the locations of ALL objects (at the instance level) in the matrix.
STRICTLY follow this exact JSON format for your response:
{"grid": 10x10 matrix (list of lists), "legend": {0: "empty", 1: "object 1", 2: "object 2", ...}}
|
[
"A. tv",
"B. fireplace",
"C. table",
"D. chair"
] |
table
|
C
|
.mov
|
241009
| null | null | null | null | null |
1,376 |
arkitscenes
|
42898849
|
object_rel_distance
|
stove
|
[Task]
This video captures an indoor scene. Your task is to first understand the spatial layout of the scene and then represent it using a 10x10 matrix.
[Rule]
1. We provide the cared categories and their corresponding instances: 1 stove, 2 sofa, 1 fireplace, 8 chair, 4 table in this scene. Focus ONLY on these categories.
2. You should use 0 to represent an empty grid and different numbers to represent different instances.
3. If a category contains multiple instances, include all of them in the matrix.
4. If objects overlap in the same grid, retain only the object with the largest area.
[Output]
Fill in the locations of ALL objects (at the instance level) in the matrix.
STRICTLY follow this exact JSON format for your response:
{"grid": 10x10 matrix (list of lists), "legend": {0: "empty", 1: "object 1", 2: "object 2", ...}}
|
[
"A. sofa",
"B. fireplace",
"C. chair",
"D. table"
] |
table
|
D
|
.mov
|
241009
| null | null | null | null | null |
1,378 |
arkitscenes
|
42899461
|
object_rel_distance
|
fireplace
|
[Task]
This video captures an indoor scene. Your task is to first understand the spatial layout of the scene and then represent it using a 10x10 matrix.
[Rule]
1. We provide the cared categories and their corresponding instances: 1 fireplace, 4 chair, 1 sofa, 1 stove, 1 tv in this scene. Focus ONLY on these categories.
2. You should use 0 to represent an empty grid and different numbers to represent different instances.
3. If a category contains multiple instances, include all of them in the matrix.
4. If objects overlap in the same grid, retain only the object with the largest area.
[Output]
Fill in the locations of ALL objects (at the instance level) in the matrix.
STRICTLY follow this exact JSON format for your response:
{"grid": 10x10 matrix (list of lists), "legend": {0: "empty", 1: "object 1", 2: "object 2", ...}}
|
[
"A. chair",
"B. sofa",
"C. stove",
"D. tv"
] |
sofa
|
B
|
.mov
|
241009
| null | null | null | null | null |
1,381 |
arkitscenes
|
42899461
|
object_rel_distance
|
fireplace
|
[Task]
This video captures an indoor scene. Your task is to first understand the spatial layout of the scene and then represent it using a 10x10 matrix.
[Rule]
1. We provide the cared categories and their corresponding instances: 1 fireplace, 3 table, 1 tv, 1 sofa, 1 stove in this scene. Focus ONLY on these categories.
2. You should use 0 to represent an empty grid and different numbers to represent different instances.
3. If a category contains multiple instances, include all of them in the matrix.
4. If objects overlap in the same grid, retain only the object with the largest area.
[Output]
Fill in the locations of ALL objects (at the instance level) in the matrix.
STRICTLY follow this exact JSON format for your response:
{"grid": 10x10 matrix (list of lists), "legend": {0: "empty", 1: "object 1", 2: "object 2", ...}}
|
[
"A. table",
"B. tv",
"C. sofa",
"D. stove"
] |
table
|
A
|
.mov
|
241009
| null | null | null | null | null |
1,395 |
arkitscenes
|
44358499
|
object_rel_distance
|
tv
|
[Task]
This video captures an indoor scene. Your task is to first understand the spatial layout of the scene and then represent it using a 10x10 matrix.
[Rule]
1. We provide the cared categories and their corresponding instances: 1 tv, 1 dishwasher, 1 fireplace, 2 sofa, 4 chair in this scene. Focus ONLY on these categories.
2. You should use 0 to represent an empty grid and different numbers to represent different instances.
3. If a category contains multiple instances, include all of them in the matrix.
4. If objects overlap in the same grid, retain only the object with the largest area.
[Output]
Fill in the locations of ALL objects (at the instance level) in the matrix.
STRICTLY follow this exact JSON format for your response:
{"grid": 10x10 matrix (list of lists), "legend": {0: "empty", 1: "object 1", 2: "object 2", ...}}
|
[
"A. dishwasher",
"B. fireplace",
"C. sofa",
"D. chair"
] |
fireplace
|
B
|
.mov
|
241009
| null | null | null | null | null |
1,405 |
arkitscenes
|
45260928
|
object_rel_distance
|
fireplace
|
[Task]
This video captures an indoor scene. Your task is to first understand the spatial layout of the scene and then represent it using a 10x10 matrix.
[Rule]
1. We provide the cared categories and their corresponding instances: 1 fireplace, 1 sofa, 1 stool, 2 chair, 2 table in this scene. Focus ONLY on these categories.
2. You should use 0 to represent an empty grid and different numbers to represent different instances.
3. If a category contains multiple instances, include all of them in the matrix.
4. If objects overlap in the same grid, retain only the object with the largest area.
[Output]
Fill in the locations of ALL objects (at the instance level) in the matrix.
STRICTLY follow this exact JSON format for your response:
{"grid": 10x10 matrix (list of lists), "legend": {0: "empty", 1: "object 1", 2: "object 2", ...}}
|
[
"A. sofa",
"B. stool",
"C. chair",
"D. table"
] |
table
|
D
|
.mov
|
241009
| null | null | null | null | null |
1,417 |
arkitscenes
|
45662924
|
object_rel_distance
|
fireplace
|
[Task]
This video captures an indoor scene. Your task is to first understand the spatial layout of the scene and then represent it using a 10x10 matrix.
[Rule]
1. We provide the cared categories and their corresponding instances: 1 fireplace, 1 table, 6 chair, 1 stove, 4 stool in this scene. Focus ONLY on these categories.
2. You should use 0 to represent an empty grid and different numbers to represent different instances.
3. If a category contains multiple instances, include all of them in the matrix.
4. If objects overlap in the same grid, retain only the object with the largest area.
[Output]
Fill in the locations of ALL objects (at the instance level) in the matrix.
STRICTLY follow this exact JSON format for your response:
{"grid": 10x10 matrix (list of lists), "legend": {0: "empty", 1: "object 1", 2: "object 2", ...}}
|
[
"A. table",
"B. chair",
"C. stove",
"D. stool"
] |
stool
|
D
|
.mov
|
241009
| null | null | null | null | null |
1,447 |
arkitscenes
|
47333457
|
object_rel_distance
|
tv
|
[Task]
This video captures an indoor scene. Your task is to first understand the spatial layout of the scene and then represent it using a 10x10 matrix.
[Rule]
1. We provide the cared categories and their corresponding instances: 1 tv, 1 toilet, 1 bed, 3 table, 1 stool in this scene. Focus ONLY on these categories.
2. You should use 0 to represent an empty grid and different numbers to represent different instances.
3. If a category contains multiple instances, include all of them in the matrix.
4. If objects overlap in the same grid, retain only the object with the largest area.
[Output]
Fill in the locations of ALL objects (at the instance level) in the matrix.
STRICTLY follow this exact JSON format for your response:
{"grid": 10x10 matrix (list of lists), "legend": {0: "empty", 1: "object 1", 2: "object 2", ...}}
|
[
"A. toilet",
"B. bed",
"C. table",
"D. stool"
] |
table
|
C
|
.mov
|
241009
| null | null | null | null | null |
1,462 |
arkitscenes
|
47334096
|
object_rel_distance
|
tv
|
[Task]
This video captures an indoor scene. Your task is to first understand the spatial layout of the scene and then represent it using a 10x10 matrix.
[Rule]
1. We provide the cared categories and their corresponding instances: 1 tv, 2 table, 1 stool, 1 sofa, 1 stove in this scene. Focus ONLY on these categories.
2. You should use 0 to represent an empty grid and different numbers to represent different instances.
3. If a category contains multiple instances, include all of them in the matrix.
4. If objects overlap in the same grid, retain only the object with the largest area.
[Output]
Fill in the locations of ALL objects (at the instance level) in the matrix.
STRICTLY follow this exact JSON format for your response:
{"grid": 10x10 matrix (list of lists), "legend": {0: "empty", 1: "object 1", 2: "object 2", ...}}
|
[
"A. table",
"B. stool",
"C. sofa",
"D. stove"
] |
table
|
A
|
.mov
|
241009
| null | null | null | null | null |
1,463 |
arkitscenes
|
47334096
|
object_rel_distance
|
stove
|
[Task]
This video captures an indoor scene. Your task is to first understand the spatial layout of the scene and then represent it using a 10x10 matrix.
[Rule]
1. We provide the cared categories and their corresponding instances: 1 stove, 1 tv, 5 chair, 1 sofa, 1 stool in this scene. Focus ONLY on these categories.
2. You should use 0 to represent an empty grid and different numbers to represent different instances.
3. If a category contains multiple instances, include all of them in the matrix.
4. If objects overlap in the same grid, retain only the object with the largest area.
[Output]
Fill in the locations of ALL objects (at the instance level) in the matrix.
STRICTLY follow this exact JSON format for your response:
{"grid": 10x10 matrix (list of lists), "legend": {0: "empty", 1: "object 1", 2: "object 2", ...}}
|
[
"A. tv",
"B. chair",
"C. sofa",
"D. stool"
] |
chair
|
B
|
.mov
|
241009
| null | null | null | null | null |
1,471 |
arkitscenes
|
47334103
|
object_rel_distance
|
tv
|
[Task]
This video captures an indoor scene. Your task is to first understand the spatial layout of the scene and then represent it using a 10x10 matrix.
[Rule]
1. We provide the cared categories and their corresponding instances: 1 tv, 1 sofa, 1 bathtub, 5 chair, 1 table in this scene. Focus ONLY on these categories.
2. You should use 0 to represent an empty grid and different numbers to represent different instances.
3. If a category contains multiple instances, include all of them in the matrix.
4. If objects overlap in the same grid, retain only the object with the largest area.
[Output]
Fill in the locations of ALL objects (at the instance level) in the matrix.
STRICTLY follow this exact JSON format for your response:
{"grid": 10x10 matrix (list of lists), "legend": {0: "empty", 1: "object 1", 2: "object 2", ...}}
|
[
"A. sofa",
"B. bathtub",
"C. chair",
"D. table"
] |
chair
|
C
|
.mov
|
241009
| null | null | null | null | null |
1,472 |
arkitscenes
|
47334103
|
object_rel_distance
|
tv
|
[Task]
This video captures an indoor scene. Your task is to first understand the spatial layout of the scene and then represent it using a 10x10 matrix.
[Rule]
1. We provide the cared categories and their corresponding instances: 1 tv, 1 sofa, 5 chair, 1 stool, 1 bathtub in this scene. Focus ONLY on these categories.
2. You should use 0 to represent an empty grid and different numbers to represent different instances.
3. If a category contains multiple instances, include all of them in the matrix.
4. If objects overlap in the same grid, retain only the object with the largest area.
[Output]
Fill in the locations of ALL objects (at the instance level) in the matrix.
STRICTLY follow this exact JSON format for your response:
{"grid": 10x10 matrix (list of lists), "legend": {0: "empty", 1: "object 1", 2: "object 2", ...}}
|
[
"A. sofa",
"B. chair",
"C. stool",
"D. bathtub"
] |
chair
|
B
|
.mov
|
241009
| null | null | null | null | null |
1,479 |
arkitscenes
|
47429977
|
object_rel_distance
|
refrigerator
|
[Task]
This video captures an indoor scene. Your task is to first understand the spatial layout of the scene and then represent it using a 10x10 matrix.
[Rule]
1. We provide the cared categories and their corresponding instances: 1 refrigerator, 1 stove, 1 tv, 2 sofa, 3 chair in this scene. Focus ONLY on these categories.
2. You should use 0 to represent an empty grid and different numbers to represent different instances.
3. If a category contains multiple instances, include all of them in the matrix.
4. If objects overlap in the same grid, retain only the object with the largest area.
[Output]
Fill in the locations of ALL objects (at the instance level) in the matrix.
STRICTLY follow this exact JSON format for your response:
{"grid": 10x10 matrix (list of lists), "legend": {0: "empty", 1: "object 1", 2: "object 2", ...}}
|
[
"A. stove",
"B. tv",
"C. sofa",
"D. chair"
] |
stove
|
A
|
.mov
|
241009
| null | null | null | null | null |
1,485 |
arkitscenes
|
47429977
|
object_rel_distance
|
tv
|
[Task]
This video captures an indoor scene. Your task is to first understand the spatial layout of the scene and then represent it using a 10x10 matrix.
[Rule]
1. We provide the cared categories and their corresponding instances: 1 tv, 1 stove, 3 chair, 1 refrigerator, 2 table in this scene. Focus ONLY on these categories.
2. You should use 0 to represent an empty grid and different numbers to represent different instances.
3. If a category contains multiple instances, include all of them in the matrix.
4. If objects overlap in the same grid, retain only the object with the largest area.
[Output]
Fill in the locations of ALL objects (at the instance level) in the matrix.
STRICTLY follow this exact JSON format for your response:
{"grid": 10x10 matrix (list of lists), "legend": {0: "empty", 1: "object 1", 2: "object 2", ...}}
|
[
"A. stove",
"B. chair",
"C. refrigerator",
"D. table"
] |
table
|
D
|
.mov
|
241009
| null | null | null | null | null |
1,500 |
arkitscenes
|
47430468
|
object_rel_distance
|
tv
|
[Task]
This video captures an indoor scene. Your task is to first understand the spatial layout of the scene and then represent it using a 10x10 matrix.
[Rule]
1. We provide the cared categories and their corresponding instances: 1 tv, 4 chair, 3 sofa, 1 refrigerator, 1 stove in this scene. Focus ONLY on these categories.
2. You should use 0 to represent an empty grid and different numbers to represent different instances.
3. If a category contains multiple instances, include all of them in the matrix.
4. If objects overlap in the same grid, retain only the object with the largest area.
[Output]
Fill in the locations of ALL objects (at the instance level) in the matrix.
STRICTLY follow this exact JSON format for your response:
{"grid": 10x10 matrix (list of lists), "legend": {0: "empty", 1: "object 1", 2: "object 2", ...}}
|
[
"A. chair",
"B. sofa",
"C. refrigerator",
"D. stove"
] |
sofa
|
B
|
.mov
|
241009
| null | null | null | null | null |
1,933 |
scannetpp
|
cc5237fd77
|
object_rel_distance
|
laptop
|
[Task]
This video captures an indoor scene. Your task is to first understand the spatial layout of the scene and then represent it using a 10x10 matrix.
[Rule]
1. We provide the cared categories and their corresponding instances: 1 laptop, 2 computer tower, 2 chair, 2 bookshelf, 2 door in this scene. Focus ONLY on these categories.
2. You should use 0 to represent an empty grid and different numbers to represent different instances.
3. If a category contains multiple instances, include all of them in the matrix.
4. If objects overlap in the same grid, retain only the object with the largest area.
[Output]
Fill in the locations of ALL objects (at the instance level) in the matrix.
STRICTLY follow this exact JSON format for your response:
{"grid": 10x10 matrix (list of lists), "legend": {0: "empty", 1: "object 1", 2: "object 2", ...}}
|
[
"A. computer tower",
"B. chair",
"C. bookshelf",
"D. door"
] |
chair
|
B
|
.mp4
|
241009
| null | null | null | null | null |
2,019 |
scannetpp
|
38d58a7a31
|
object_rel_distance
|
keyboard
|
[Task]
This video captures an indoor scene. Your task is to first understand the spatial layout of the scene and then represent it using a 10x10 matrix.
[Rule]
1. We provide the cared categories and their corresponding instances: 1 keyboard, 1 whiteboard, 10 ceiling light, 3 heater, 2 coat hanger in this scene. Focus ONLY on these categories.
2. You should use 0 to represent an empty grid and different numbers to represent different instances.
3. If a category contains multiple instances, include all of them in the matrix.
4. If objects overlap in the same grid, retain only the object with the largest area.
[Output]
Fill in the locations of ALL objects (at the instance level) in the matrix.
STRICTLY follow this exact JSON format for your response:
{"grid": 10x10 matrix (list of lists), "legend": {0: "empty", 1: "object 1", 2: "object 2", ...}}
|
[
"A. whiteboard",
"B. ceiling light",
"C. heater",
"D. coat hanger"
] |
heater
|
C
|
.mp4
|
241009
| null | null | null | null | null |
2,023 |
scannetpp
|
38d58a7a31
|
object_rel_distance
|
telephone
|
[Task]
This video captures an indoor scene. Your task is to first understand the spatial layout of the scene and then represent it using a 10x10 matrix.
[Rule]
1. We provide the cared categories and their corresponding instances: 1 telephone, 2 coat hanger, 1 trash can, 2 door, 8 table in this scene. Focus ONLY on these categories.
2. You should use 0 to represent an empty grid and different numbers to represent different instances.
3. If a category contains multiple instances, include all of them in the matrix.
4. If objects overlap in the same grid, retain only the object with the largest area.
[Output]
Fill in the locations of ALL objects (at the instance level) in the matrix.
STRICTLY follow this exact JSON format for your response:
{"grid": 10x10 matrix (list of lists), "legend": {0: "empty", 1: "object 1", 2: "object 2", ...}}
|
[
"A. coat hanger",
"B. trash can",
"C. door",
"D. table"
] |
table
|
D
|
.mp4
|
241009
| null | null | null | null | null |
2,051 |
scannetpp
|
25f3b7a318
|
object_rel_distance
|
door
|
[Task]
This video captures an indoor scene. Your task is to first understand the spatial layout of the scene and then represent it using a 10x10 matrix.
[Rule]
1. We provide the cared categories and their corresponding instances: 1 door, 1 blanket, 1 laptop, 2 chair, 1 basket in this scene. Focus ONLY on these categories.
2. You should use 0 to represent an empty grid and different numbers to represent different instances.
3. If a category contains multiple instances, include all of them in the matrix.
4. If objects overlap in the same grid, retain only the object with the largest area.
[Output]
Fill in the locations of ALL objects (at the instance level) in the matrix.
STRICTLY follow this exact JSON format for your response:
{"grid": 10x10 matrix (list of lists), "legend": {0: "empty", 1: "object 1", 2: "object 2", ...}}
|
[
"A. blanket",
"B. laptop",
"C. chair",
"D. basket"
] |
basket
|
D
|
.mp4
|
241009
| null | null | null | null | null |
2,060 |
scannetpp
|
bcd2436daf
|
object_rel_distance
|
pillow
|
[Task]
This video captures an indoor scene. Your task is to first understand the spatial layout of the scene and then represent it using a 10x10 matrix.
[Rule]
1. We provide the cared categories and their corresponding instances: 1 pillow, 1 heater, 1 plant, 1 chair, 1 ceiling light in this scene. Focus ONLY on these categories.
2. You should use 0 to represent an empty grid and different numbers to represent different instances.
3. If a category contains multiple instances, include all of them in the matrix.
4. If objects overlap in the same grid, retain only the object with the largest area.
[Output]
Fill in the locations of ALL objects (at the instance level) in the matrix.
STRICTLY follow this exact JSON format for your response:
{"grid": 10x10 matrix (list of lists), "legend": {0: "empty", 1: "object 1", 2: "object 2", ...}}
|
[
"A. heater",
"B. plant",
"C. chair",
"D. ceiling light"
] |
plant
|
B
|
.mp4
|
241009
| null | null | null | null | null |
2,073 |
scannetpp
|
0d2ee665be
|
object_rel_distance
|
trash can
|
[Task]
This video captures an indoor scene. Your task is to first understand the spatial layout of the scene and then represent it using a 10x10 matrix.
[Rule]
1. We provide the cared categories and their corresponding instances: 1 trash can, 1 microwave, 1 bed, 1 laptop, 2 chair in this scene. Focus ONLY on these categories.
2. You should use 0 to represent an empty grid and different numbers to represent different instances.
3. If a category contains multiple instances, include all of them in the matrix.
4. If objects overlap in the same grid, retain only the object with the largest area.
[Output]
Fill in the locations of ALL objects (at the instance level) in the matrix.
STRICTLY follow this exact JSON format for your response:
{"grid": 10x10 matrix (list of lists), "legend": {0: "empty", 1: "object 1", 2: "object 2", ...}}
|
[
"A. microwave",
"B. bed",
"C. laptop",
"D. chair"
] |
laptop
|
C
|
.mp4
|
241009
| null | null | null | null | null |
2,093 |
scannetpp
|
ac48a9b736
|
object_rel_distance
|
trash can
|
[Task]
This video captures an indoor scene. Your task is to first understand the spatial layout of the scene and then represent it using a 10x10 matrix.
[Rule]
1. We provide the cared categories and their corresponding instances: 1 trash can, 3 door, 9 chair, 8 table, 6 keyboard in this scene. Focus ONLY on these categories.
2. You should use 0 to represent an empty grid and different numbers to represent different instances.
3. If a category contains multiple instances, include all of them in the matrix.
4. If objects overlap in the same grid, retain only the object with the largest area.
[Output]
Fill in the locations of ALL objects (at the instance level) in the matrix.
STRICTLY follow this exact JSON format for your response:
{"grid": 10x10 matrix (list of lists), "legend": {0: "empty", 1: "object 1", 2: "object 2", ...}}
|
[
"A. door",
"B. chair",
"C. table",
"D. keyboard"
] |
door
|
A
|
.mp4
|
241009
| null | null | null | null | null |
2,101 |
scannetpp
|
ac48a9b736
|
object_rel_distance
|
whiteboard
|
[Task]
This video captures an indoor scene. Your task is to first understand the spatial layout of the scene and then represent it using a 10x10 matrix.
[Rule]
1. We provide the cared categories and their corresponding instances: 1 whiteboard, 3 door, 9 monitor, 8 table, 2 computer mouse in this scene. Focus ONLY on these categories.
2. You should use 0 to represent an empty grid and different numbers to represent different instances.
3. If a category contains multiple instances, include all of them in the matrix.
4. If objects overlap in the same grid, retain only the object with the largest area.
[Output]
Fill in the locations of ALL objects (at the instance level) in the matrix.
STRICTLY follow this exact JSON format for your response:
{"grid": 10x10 matrix (list of lists), "legend": {0: "empty", 1: "object 1", 2: "object 2", ...}}
|
[
"A. door",
"B. monitor",
"C. table",
"D. computer mouse"
] |
table
|
C
|
.mp4
|
241009
| null | null | null | null | null |
2,107 |
scannetpp
|
578511c8a9
|
object_rel_distance
|
printer
|
[Task]
This video captures an indoor scene. Your task is to first understand the spatial layout of the scene and then represent it using a 10x10 matrix.
[Rule]
1. We provide the cared categories and their corresponding instances: 1 printer, 1 telephone, 16 monitor, 1 kettle, 1 table lamp in this scene. Focus ONLY on these categories.
2. You should use 0 to represent an empty grid and different numbers to represent different instances.
3. If a category contains multiple instances, include all of them in the matrix.
4. If objects overlap in the same grid, retain only the object with the largest area.
[Output]
Fill in the locations of ALL objects (at the instance level) in the matrix.
STRICTLY follow this exact JSON format for your response:
{"grid": 10x10 matrix (list of lists), "legend": {0: "empty", 1: "object 1", 2: "object 2", ...}}
|
[
"A. telephone",
"B. monitor",
"C. kettle",
"D. table lamp"
] |
kettle
|
C
|
.mp4
|
241009
| null | null | null | null | null |
2,136 |
scannetpp
|
09c1414f1b
|
object_rel_distance
|
microwave
|
[Task]
This video captures an indoor scene. Your task is to first understand the spatial layout of the scene and then represent it using a 10x10 matrix.
[Rule]
1. We provide the cared categories and their corresponding instances: 1 microwave, 1 tv, 1 trash can, 2 pillow, 1 cutting board in this scene. Focus ONLY on these categories.
2. You should use 0 to represent an empty grid and different numbers to represent different instances.
3. If a category contains multiple instances, include all of them in the matrix.
4. If objects overlap in the same grid, retain only the object with the largest area.
[Output]
Fill in the locations of ALL objects (at the instance level) in the matrix.
STRICTLY follow this exact JSON format for your response:
{"grid": 10x10 matrix (list of lists), "legend": {0: "empty", 1: "object 1", 2: "object 2", ...}}
|
[
"A. tv",
"B. trash can",
"C. pillow",
"D. cutting board"
] |
trash can
|
B
|
.mp4
|
241009
| null | null | null | null | null |
2,140 |
scannetpp
|
5f99900f09
|
object_rel_distance
|
suitcase
|
[Task]
This video captures an indoor scene. Your task is to first understand the spatial layout of the scene and then represent it using a 10x10 matrix.
[Rule]
1. We provide the cared categories and their corresponding instances: 1 suitcase, 2 door, 5 ceiling light, 1 refrigerator, 2 monitor in this scene. Focus ONLY on these categories.
2. You should use 0 to represent an empty grid and different numbers to represent different instances.
3. If a category contains multiple instances, include all of them in the matrix.
4. If objects overlap in the same grid, retain only the object with the largest area.
[Output]
Fill in the locations of ALL objects (at the instance level) in the matrix.
STRICTLY follow this exact JSON format for your response:
{"grid": 10x10 matrix (list of lists), "legend": {0: "empty", 1: "object 1", 2: "object 2", ...}}
|
[
"A. door",
"B. ceiling light",
"C. refrigerator",
"D. monitor"
] |
door
|
A
|
.mp4
|
241009
| null | null | null | null | null |
2,145 |
scannetpp
|
5f99900f09
|
object_rel_distance
|
suitcase
|
[Task]
This video captures an indoor scene. Your task is to first understand the spatial layout of the scene and then represent it using a 10x10 matrix.
[Rule]
1. We provide the cared categories and their corresponding instances: 1 suitcase, 2 printer, 1 refrigerator, 11 table, 14 chair in this scene. Focus ONLY on these categories.
2. You should use 0 to represent an empty grid and different numbers to represent different instances.
3. If a category contains multiple instances, include all of them in the matrix.
4. If objects overlap in the same grid, retain only the object with the largest area.
[Output]
Fill in the locations of ALL objects (at the instance level) in the matrix.
STRICTLY follow this exact JSON format for your response:
{"grid": 10x10 matrix (list of lists), "legend": {0: "empty", 1: "object 1", 2: "object 2", ...}}
|
[
"A. printer",
"B. refrigerator",
"C. table",
"D. chair"
] |
refrigerator
|
B
|
.mp4
|
241009
| null | null | null | null | null |
2,151 |
scannetpp
|
5f99900f09
|
object_rel_distance
|
suitcase
|
[Task]
This video captures an indoor scene. Your task is to first understand the spatial layout of the scene and then represent it using a 10x10 matrix.
[Rule]
1. We provide the cared categories and their corresponding instances: 1 suitcase, 2 printer, 14 chair, 3 whiteboard, 11 table in this scene. Focus ONLY on these categories.
2. You should use 0 to represent an empty grid and different numbers to represent different instances.
3. If a category contains multiple instances, include all of them in the matrix.
4. If objects overlap in the same grid, retain only the object with the largest area.
[Output]
Fill in the locations of ALL objects (at the instance level) in the matrix.
STRICTLY follow this exact JSON format for your response:
{"grid": 10x10 matrix (list of lists), "legend": {0: "empty", 1: "object 1", 2: "object 2", ...}}
|
[
"A. printer",
"B. chair",
"C. whiteboard",
"D. table"
] |
chair
|
B
|
.mp4
|
241009
| null | null | null | null | null |
2,154 |
scannetpp
|
5f99900f09
|
object_rel_distance
|
refrigerator
|
[Task]
This video captures an indoor scene. Your task is to first understand the spatial layout of the scene and then represent it using a 10x10 matrix.
[Rule]
1. We provide the cared categories and their corresponding instances: 1 refrigerator, 2 door, 3 whiteboard, 14 chair, 11 table in this scene. Focus ONLY on these categories.
2. You should use 0 to represent an empty grid and different numbers to represent different instances.
3. If a category contains multiple instances, include all of them in the matrix.
4. If objects overlap in the same grid, retain only the object with the largest area.
[Output]
Fill in the locations of ALL objects (at the instance level) in the matrix.
STRICTLY follow this exact JSON format for your response:
{"grid": 10x10 matrix (list of lists), "legend": {0: "empty", 1: "object 1", 2: "object 2", ...}}
|
[
"A. door",
"B. whiteboard",
"C. chair",
"D. table"
] |
chair
|
C
|
.mp4
|
241009
| null | null | null | null | null |
2,969 |
scannet
|
scene0356_00
|
object_rel_distance
|
telephone
|
[Task]
This video captures an indoor scene. Your task is to first understand the spatial layout of the scene and then represent it using a 10x10 matrix.
[Rule]
1. We provide the cared categories and their corresponding instances: 1 telephone, 1 table, 2 trash bin, 1 bed, 1 window in this scene. Focus ONLY on these categories.
2. You should use 0 to represent an empty grid and different numbers to represent different instances.
3. If a category contains multiple instances, include all of them in the matrix.
4. If objects overlap in the same grid, retain only the object with the largest area.
[Output]
Fill in the locations of ALL objects (at the instance level) in the matrix.
STRICTLY follow this exact JSON format for your response:
{"grid": 10x10 matrix (list of lists), "legend": {0: "empty", 1: "object 1", 2: "object 2", ...}}
|
[
"A. table",
"B. trash bin",
"C. bed",
"D. window"
] |
bed
|
C
|
.mp4
|
241009
| null | null | null | null | null |
2,975 |
scannet
|
scene0378_01
|
object_rel_distance
|
monitor
|
[Task]
This video captures an indoor scene. Your task is to first understand the spatial layout of the scene and then represent it using a 10x10 matrix.
[Rule]
1. We provide the cared categories and their corresponding instances: 1 monitor, 1 clock, 2 bookshelf, 1 door, 4 chair in this scene. Focus ONLY on these categories.
2. You should use 0 to represent an empty grid and different numbers to represent different instances.
3. If a category contains multiple instances, include all of them in the matrix.
4. If objects overlap in the same grid, retain only the object with the largest area.
[Output]
Fill in the locations of ALL objects (at the instance level) in the matrix.
STRICTLY follow this exact JSON format for your response:
{"grid": 10x10 matrix (list of lists), "legend": {0: "empty", 1: "object 1", 2: "object 2", ...}}
|
[
"A. clock",
"B. bookshelf",
"C. door",
"D. chair"
] |
chair
|
D
|
.mp4
|
241009
| null | null | null | null | null |
2,978 |
scannet
|
scene0378_01
|
object_rel_distance
|
door
|
[Task]
This video captures an indoor scene. Your task is to first understand the spatial layout of the scene and then represent it using a 10x10 matrix.
[Rule]
1. We provide the cared categories and their corresponding instances: 1 door, 1 window, 3 table, 1 trash bin, 1 printer in this scene. Focus ONLY on these categories.
2. You should use 0 to represent an empty grid and different numbers to represent different instances.
3. If a category contains multiple instances, include all of them in the matrix.
4. If objects overlap in the same grid, retain only the object with the largest area.
[Output]
Fill in the locations of ALL objects (at the instance level) in the matrix.
STRICTLY follow this exact JSON format for your response:
{"grid": 10x10 matrix (list of lists), "legend": {0: "empty", 1: "object 1", 2: "object 2", ...}}
|
[
"A. window",
"B. table",
"C. trash bin",
"D. printer"
] |
table
|
B
|
.mp4
|
241009
| null | null | null | null | null |
2,985 |
scannet
|
scene0426_00
|
object_rel_distance
|
window
|
[Task]
This video captures an indoor scene. Your task is to first understand the spatial layout of the scene and then represent it using a 10x10 matrix.
[Rule]
1. We provide the cared categories and their corresponding instances: 1 window, 3 chair, 1 lamp, 1 tv, 1 door in this scene. Focus ONLY on these categories.
2. You should use 0 to represent an empty grid and different numbers to represent different instances.
3. If a category contains multiple instances, include all of them in the matrix.
4. If objects overlap in the same grid, retain only the object with the largest area.
[Output]
Fill in the locations of ALL objects (at the instance level) in the matrix.
STRICTLY follow this exact JSON format for your response:
{"grid": 10x10 matrix (list of lists), "legend": {0: "empty", 1: "object 1", 2: "object 2", ...}}
|
[
"A. chair",
"B. lamp",
"C. tv",
"D. door"
] |
chair
|
A
|
.mp4
|
241009
| null | null | null | null | null |
3,016 |
scannet
|
scene0647_00
|
object_rel_distance
|
tv
|
[Task]
This video captures an indoor scene. Your task is to first understand the spatial layout of the scene and then represent it using a 10x10 matrix.
[Rule]
1. We provide the cared categories and their corresponding instances: 1 tv, 2 door, 3 table, 1 chair, 1 backpack in this scene. Focus ONLY on these categories.
2. You should use 0 to represent an empty grid and different numbers to represent different instances.
3. If a category contains multiple instances, include all of them in the matrix.
4. If objects overlap in the same grid, retain only the object with the largest area.
[Output]
Fill in the locations of ALL objects (at the instance level) in the matrix.
STRICTLY follow this exact JSON format for your response:
{"grid": 10x10 matrix (list of lists), "legend": {0: "empty", 1: "object 1", 2: "object 2", ...}}
|
[
"A. door",
"B. table",
"C. chair",
"D. backpack"
] |
table
|
B
|
.mp4
|
241009
| null | null | null | null | null |
3,027 |
scannet
|
scene0651_02
|
object_rel_distance
|
trash bin
|
[Task]
This video captures an indoor scene. Your task is to first understand the spatial layout of the scene and then represent it using a 10x10 matrix.
[Rule]
1. We provide the cared categories and their corresponding instances: 1 trash bin, 2 table, 1 sofa, 1 counter, 4 chair in this scene. Focus ONLY on these categories.
2. You should use 0 to represent an empty grid and different numbers to represent different instances.
3. If a category contains multiple instances, include all of them in the matrix.
4. If objects overlap in the same grid, retain only the object with the largest area.
[Output]
Fill in the locations of ALL objects (at the instance level) in the matrix.
STRICTLY follow this exact JSON format for your response:
{"grid": 10x10 matrix (list of lists), "legend": {0: "empty", 1: "object 1", 2: "object 2", ...}}
|
[
"A. table",
"B. sofa",
"C. counter",
"D. chair"
] |
counter
|
C
|
.mp4
|
241009
| null | null | null | null | null |
3,046 |
scannet
|
scene0653_00
|
object_rel_distance
|
door
|
[Task]
This video captures an indoor scene. Your task is to first understand the spatial layout of the scene and then represent it using a 10x10 matrix.
[Rule]
1. We provide the cared categories and their corresponding instances: 1 door, 7 monitor, 2 window, 2 trash bin, 8 chair in this scene. Focus ONLY on these categories.
2. You should use 0 to represent an empty grid and different numbers to represent different instances.
3. If a category contains multiple instances, include all of them in the matrix.
4. If objects overlap in the same grid, retain only the object with the largest area.
[Output]
Fill in the locations of ALL objects (at the instance level) in the matrix.
STRICTLY follow this exact JSON format for your response:
{"grid": 10x10 matrix (list of lists), "legend": {0: "empty", 1: "object 1", 2: "object 2", ...}}
|
[
"A. monitor",
"B. window",
"C. trash bin",
"D. chair"
] |
monitor
|
A
|
.mp4
|
241009
| null | null | null | null | null |
3,070 |
scannet
|
scene0697_01
|
object_rel_distance
|
closet
|
[Task]
This video captures an indoor scene. Your task is to first understand the spatial layout of the scene and then represent it using a 10x10 matrix.
[Rule]
1. We provide the cared categories and their corresponding instances: 1 closet, 2 bed, 1 tv, 1 lamp, 1 mirror in this scene. Focus ONLY on these categories.
2. You should use 0 to represent an empty grid and different numbers to represent different instances.
3. If a category contains multiple instances, include all of them in the matrix.
4. If objects overlap in the same grid, retain only the object with the largest area.
[Output]
Fill in the locations of ALL objects (at the instance level) in the matrix.
STRICTLY follow this exact JSON format for your response:
{"grid": 10x10 matrix (list of lists), "legend": {0: "empty", 1: "object 1", 2: "object 2", ...}}
|
[
"A. bed",
"B. tv",
"C. lamp",
"D. mirror"
] |
bed
|
A
|
.mp4
|
241009
| null | null | null | null | null |
3,072 |
scannet
|
scene0697_01
|
object_rel_distance
|
closet
|
[Task]
This video captures an indoor scene. Your task is to first understand the spatial layout of the scene and then represent it using a 10x10 matrix.
[Rule]
1. We provide the cared categories and their corresponding instances: 1 closet, 2 window, 1 trash bin, 1 mirror, 3 pillow in this scene. Focus ONLY on these categories.
2. You should use 0 to represent an empty grid and different numbers to represent different instances.
3. If a category contains multiple instances, include all of them in the matrix.
4. If objects overlap in the same grid, retain only the object with the largest area.
[Output]
Fill in the locations of ALL objects (at the instance level) in the matrix.
STRICTLY follow this exact JSON format for your response:
{"grid": 10x10 matrix (list of lists), "legend": {0: "empty", 1: "object 1", 2: "object 2", ...}}
|
[
"A. window",
"B. trash bin",
"C. mirror",
"D. pillow"
] |
trash bin
|
B
|
.mp4
|
241009
| null | null | null | null | null |
3,075 |
scannet
|
scene0697_01
|
object_rel_distance
|
closet
|
[Task]
This video captures an indoor scene. Your task is to first understand the spatial layout of the scene and then represent it using a 10x10 matrix.
[Rule]
1. We provide the cared categories and their corresponding instances: 1 closet, 1 lamp, 2 bed, 1 tv, 2 window in this scene. Focus ONLY on these categories.
2. You should use 0 to represent an empty grid and different numbers to represent different instances.
3. If a category contains multiple instances, include all of them in the matrix.
4. If objects overlap in the same grid, retain only the object with the largest area.
[Output]
Fill in the locations of ALL objects (at the instance level) in the matrix.
STRICTLY follow this exact JSON format for your response:
{"grid": 10x10 matrix (list of lists), "legend": {0: "empty", 1: "object 1", 2: "object 2", ...}}
|
[
"A. lamp",
"B. bed",
"C. tv",
"D. window"
] |
bed
|
B
|
.mp4
|
241009
| null | null | null | null | null |
3,089 |
scannet
|
scene0700_02
|
object_rel_distance
|
window
|
[Task]
This video captures an indoor scene. Your task is to first understand the spatial layout of the scene and then represent it using a 10x10 matrix.
[Rule]
1. We provide the cared categories and their corresponding instances: 1 window, 1 keyboard, 2 backpack, 1 telephone, 4 chair in this scene. Focus ONLY on these categories.
2. You should use 0 to represent an empty grid and different numbers to represent different instances.
3. If a category contains multiple instances, include all of them in the matrix.
4. If objects overlap in the same grid, retain only the object with the largest area.
[Output]
Fill in the locations of ALL objects (at the instance level) in the matrix.
STRICTLY follow this exact JSON format for your response:
{"grid": 10x10 matrix (list of lists), "legend": {0: "empty", 1: "object 1", 2: "object 2", ...}}
|
[
"A. keyboard",
"B. backpack",
"C. telephone",
"D. chair"
] |
chair
|
D
|
.mp4
|
241009
| null | null | null | null | null |
3,102 |
scannet
|
scene0458_00
|
object_rel_distance
|
mirror
|
[Task]
This video captures an indoor scene. Your task is to first understand the spatial layout of the scene and then represent it using a 10x10 matrix.
[Rule]
1. We provide the cared categories and their corresponding instances: 1 mirror, 2 trash bin, 1 window, 1 towel, 1 door in this scene. Focus ONLY on these categories.
2. You should use 0 to represent an empty grid and different numbers to represent different instances.
3. If a category contains multiple instances, include all of them in the matrix.
4. If objects overlap in the same grid, retain only the object with the largest area.
[Output]
Fill in the locations of ALL objects (at the instance level) in the matrix.
STRICTLY follow this exact JSON format for your response:
{"grid": 10x10 matrix (list of lists), "legend": {0: "empty", 1: "object 1", 2: "object 2", ...}}
|
[
"A. trash bin",
"B. window",
"C. towel",
"D. door"
] |
trash bin
|
A
|
.mp4
|
241009
| null | null | null | null | null |
3,133 |
scannet
|
scene0608_00
|
object_rel_distance
|
door
|
[Task]
This video captures an indoor scene. Your task is to first understand the spatial layout of the scene and then represent it using a 10x10 matrix.
[Rule]
1. We provide the cared categories and their corresponding instances: 1 door, 1 tv, 1 window, 3 chair, 3 table in this scene. Focus ONLY on these categories.
2. You should use 0 to represent an empty grid and different numbers to represent different instances.
3. If a category contains multiple instances, include all of them in the matrix.
4. If objects overlap in the same grid, retain only the object with the largest area.
[Output]
Fill in the locations of ALL objects (at the instance level) in the matrix.
STRICTLY follow this exact JSON format for your response:
{"grid": 10x10 matrix (list of lists), "legend": {0: "empty", 1: "object 1", 2: "object 2", ...}}
|
[
"A. tv",
"B. window",
"C. chair",
"D. table"
] |
chair
|
C
|
.mp4
|
241009
| null | null | null | null | null |
3,143 |
scannet
|
scene0030_02
|
object_rel_distance
|
door
|
[Task]
This video captures an indoor scene. Your task is to first understand the spatial layout of the scene and then represent it using a 10x10 matrix.
[Rule]
1. We provide the cared categories and their corresponding instances: 1 door, 6 table, 1 clock, 17 chair, 1 trash bin in this scene. Focus ONLY on these categories.
2. You should use 0 to represent an empty grid and different numbers to represent different instances.
3. If a category contains multiple instances, include all of them in the matrix.
4. If objects overlap in the same grid, retain only the object with the largest area.
[Output]
Fill in the locations of ALL objects (at the instance level) in the matrix.
STRICTLY follow this exact JSON format for your response:
{"grid": 10x10 matrix (list of lists), "legend": {0: "empty", 1: "object 1", 2: "object 2", ...}}
|
[
"A. table",
"B. clock",
"C. chair",
"D. trash bin"
] |
trash bin
|
D
|
.mp4
|
241009
| null | null | null | null | null |
3,150 |
scannet
|
scene0046_01
|
object_rel_distance
|
lamp
|
[Task]
This video captures an indoor scene. Your task is to first understand the spatial layout of the scene and then represent it using a 10x10 matrix.
[Rule]
1. We provide the cared categories and their corresponding instances: 1 lamp, 1 chair, 2 door, 1 mirror, 2 backpack in this scene. Focus ONLY on these categories.
2. You should use 0 to represent an empty grid and different numbers to represent different instances.
3. If a category contains multiple instances, include all of them in the matrix.
4. If objects overlap in the same grid, retain only the object with the largest area.
[Output]
Fill in the locations of ALL objects (at the instance level) in the matrix.
STRICTLY follow this exact JSON format for your response:
{"grid": 10x10 matrix (list of lists), "legend": {0: "empty", 1: "object 1", 2: "object 2", ...}}
|
[
"A. chair",
"B. door",
"C. mirror",
"D. backpack"
] |
backpack
|
D
|
.mp4
|
241009
| null | null | null | null | null |
3,179 |
scannet
|
scene0207_02
|
object_rel_distance
|
counter
|
[Task]
This video captures an indoor scene. Your task is to first understand the spatial layout of the scene and then represent it using a 10x10 matrix.
[Rule]
1. We provide the cared categories and their corresponding instances: 1 counter, 1 microwave, 2 backpack, 1 nightstand, 1 closet in this scene. Focus ONLY on these categories.
2. You should use 0 to represent an empty grid and different numbers to represent different instances.
3. If a category contains multiple instances, include all of them in the matrix.
4. If objects overlap in the same grid, retain only the object with the largest area.
[Output]
Fill in the locations of ALL objects (at the instance level) in the matrix.
STRICTLY follow this exact JSON format for your response:
{"grid": 10x10 matrix (list of lists), "legend": {0: "empty", 1: "object 1", 2: "object 2", ...}}
|
[
"A. microwave",
"B. backpack",
"C. nightstand",
"D. closet"
] |
backpack
|
B
|
.mp4
|
241009
| null | null | null | null | null |
3,180 |
scannet
|
scene0207_02
|
object_rel_distance
|
pillow
|
[Task]
This video captures an indoor scene. Your task is to first understand the spatial layout of the scene and then represent it using a 10x10 matrix.
[Rule]
1. We provide the cared categories and their corresponding instances: 1 pillow, 2 door, 1 refrigerator, 1 chair, 2 table in this scene. Focus ONLY on these categories.
2. You should use 0 to represent an empty grid and different numbers to represent different instances.
3. If a category contains multiple instances, include all of them in the matrix.
4. If objects overlap in the same grid, retain only the object with the largest area.
[Output]
Fill in the locations of ALL objects (at the instance level) in the matrix.
STRICTLY follow this exact JSON format for your response:
{"grid": 10x10 matrix (list of lists), "legend": {0: "empty", 1: "object 1", 2: "object 2", ...}}
|
[
"A. door",
"B. refrigerator",
"C. chair",
"D. table"
] |
table
|
D
|
.mp4
|
241009
| null | null | null | null | null |
3,182 |
scannet
|
scene0221_01
|
object_rel_distance
|
trash bin
|
[Task]
This video captures an indoor scene. Your task is to first understand the spatial layout of the scene and then represent it using a 10x10 matrix.
[Rule]
1. We provide the cared categories and their corresponding instances: 1 trash bin, 2 bed, 2 door, 1 nightstand, 2 table in this scene. Focus ONLY on these categories.
2. You should use 0 to represent an empty grid and different numbers to represent different instances.
3. If a category contains multiple instances, include all of them in the matrix.
4. If objects overlap in the same grid, retain only the object with the largest area.
[Output]
Fill in the locations of ALL objects (at the instance level) in the matrix.
STRICTLY follow this exact JSON format for your response:
{"grid": 10x10 matrix (list of lists), "legend": {0: "empty", 1: "object 1", 2: "object 2", ...}}
|
[
"A. bed",
"B. door",
"C. nightstand",
"D. table"
] |
bed
|
A
|
.mp4
|
241009
| null | null | null | null | null |
No dataset card yet
- Downloads last month
- 3