These 50 samples showcase what the SpatialClaw agent
is doing under the hood — composing perception tools, calling into
numpy / scipy / matplotlib on a
persistent Python kernel, and revising its analysis after inspecting
intermediate results. Click any tile to see the full trajectory.
Question
Assuming that in Figure 4, the wall where the sink is located and the direction pointing outward from the wall is east, then in Figure 1, what is the position of the door relative to the sink in Figure 4? Options: A: No…
Summary
Composes 3D reconstruction with mask segmentation, decomposes via SVD. 63 cross-step variable references demonstrate state persistence across steps. 7 targeted VLM queries.
Open trajectory →
Question
The fireplace faces north; in which direction is the painting on the wall in the fitness area facing? Options: A: Southwest, B: East, C: Northwest, D: Northeast Answer with the option's letter from the given choices di…
Summary
Composes 3D reconstruction with mask segmentation, decomposes via SVD.
Open trajectory →
Question
If the right-most chair is 0.2m wide, how tall is the wooden cabinet? This is a spatial reasoning question about an indoor scene. Answer with a single value: a number (float or integer) or a short text answer. Do NOT i…
Summary
Composes 3D reconstruction with mask segmentation, reaches into
scipy.spatial. 76 cross-step variable references demonstrate state persistence across steps. 6 targeted VLM queries.Open trajectory →
Question
Measuring from the closest point of each object, what is the distance between the sofa and the toilet (in meters)? These are frames of a video. Answer the question. For multiple-choice, answer with the option's letter …
Summary
Composes 3D reconstruction with mask segmentation, reaches into
scipy.spatial. Iterative visual inspection: 12 show() calls.Open trajectory →
Question
If I am standing by the ceiling light and facing the monitor, is the door to the left or the right of the monitor? These are frames of a video. Answer the question. For multiple-choice, answer with the option's letter …
Summary
Composes 3D reconstruction with mask segmentation, plots intermediate state with
matplotlib, computes cross products.Open trajectory →
Question
Measuring from the closest point of each object, what is the distance between the computer tower and the heater (in meters)? These are frames of a video. Answer the question. For multiple-choice, answer with the option…
Summary
Composes 3D reconstruction with mask segmentation, reaches into
scipy.spatial. 6 targeted VLM queries.Open trajectory →
Question
If I am standing by the laptop and facing the chair, is the monitor to my left-front, left-back, right-front, or right-back? The directions refer to the quadrants of a Cartesian plane (if I am standing at the origin and…
Summary
Composes 3D reconstruction with mask segmentation, plots intermediate state with
matplotlib, computes cross products.Open trajectory →
Question
Imagine standing at desk looking towards television, where is toilet? A. back-left B. back-right C. front D. back This question is from ViewSpatial-Bench. The task tests spatial localization either from the camera's o…
Summary
Composes 3D reconstruction with mask segmentation, plots intermediate state with
matplotlib, computes cross products.Open trajectory →
Question
If I am standing by the monitor and facing the chair, is the laptop to my left, right, front, or back? The directions refer to the quadrants of a Cartesian plane (if I am standing at the origin and facing along the posi…
Summary
Composes 3D reconstruction with mask segmentation, plots intermediate state with
matplotlib, computes cross products.Open trajectory →
Question
The overall goal is to "Checkout and scan barcode in the supermarket". The agent in the video is currently performing one subtask out of many to complete this goal. For the agent in the video, what is the most plausible…
Summary
Composes 3D reconstruction with mask segmentation, plots intermediate state with
matplotlib.Open trajectory →
Question
Measuring from the closest point of each object, what is the distance between the radiator and the chair (in meters)? Answer the spatial reasoning question about the 3D scene. For multiple-choice, answer with the optio…
Summary
Composes 3D reconstruction with mask segmentation, reaches into
scipy.spatial.Open trajectory →
Question
Which direction is the door relative to the socket? Options: A: Front left, B: Front right, C: Rear left, D: Rear right Answer with the option's letter from the given choices directly.
Summary
Composes 3D reconstruction with mask segmentation, decomposes via SVD.
Open trajectory →
Question
If the direction the iMac screen faces is considered the front, in which direction is the blackboard relative to the iMac? Options: A: Left, B: Back, C: Right, D: Front Answer with the option's letter from the given ch…
Summary
Composes 3D reconstruction with mask segmentation, decomposes via SVD.
Open trajectory →
Question
Measuring from the closest point of each object, what is the distance between the stool and the washer (in meters)? These are frames of a video. Answer the question. For multiple-choice, answer with the option's letter…
Summary
Composes 3D reconstruction with mask segmentation, reaches into
scipy.spatial. 101 cross-step variable references demonstrate state persistence across steps. Iterative visual inspection: 18 show() calls. 17 targeted VLM queries.Open trajectory →
Question
Measuring from the closest point of each object, which of these objects (table, fireplace, bed, chair) is the closest to the tv? These are frames of a video. Answer the question. For multiple-choice, answer with the op…
Summary
Composes 3D reconstruction with mask segmentation, reaches into
scipy.spatial. 7 targeted VLM queries.Open trajectory →
Question
With the direction facing the wall from the first bulletin board as the front, in which direction is the fitness area relative to the bulletin board? Options: A: Right, B: Left, C: Back, D: Front Answer with the option…
Summary
Composes 3D reconstruction with mask segmentation, decomposes via SVD. 5 targeted VLM queries.
Open trajectory →
Question
Measuring from the closest point of each object, which of the following is closest to the green grass(id: 11): green park bench(id: 05), white signboard(id: 13), white pole(id: 08), or wooden bench(id: 09)? These are f…
Summary
Composes 3D reconstruction with mask segmentation, reaches into
scipy.spatial. 64 cross-step variable references demonstrate state persistence across steps. 5 targeted VLM queries.Open trajectory →
Question
If the toilet paper falls direftly towards the toilet, will it hit the toilet or the floor first? This is a spatial reasoning question about an indoor scene. Answer with a single value: a number (float or integer) or a…
Summary
Composes 3D reconstruction with mask segmentation.
Open trajectory →
Question
You are a robot beginning at the door and facing the lamp. You want to navigate to the bookshelf. You will perform the following actions (Note: for each [please fill in], choose either 'turn back,' 'turn left,' or 'turn…
Summary
Composes 3D reconstruction with mask segmentation, plots intermediate state with
matplotlib. 5 targeted VLM queries.Open trajectory →
Question
I'm standing at the christmas tree and facing the camera. Call this direction north (N). What direction is the TV? Choose from N, NE, E, SE, S, SW, W, NW. This is a spatial reasoning question about an indoor scene. Ans…
Summary
Composes 3D reconstruction with mask segmentation, computes cross products.
Open trajectory →
Question
In this video clip, relative to its starting orientation and location, how is the observer's location moving? This is a multiple-choice question about how an object or the camera moves in 3D space within the video clip…
Summary
Builds a 3D reconstruction across frames, computes angles with
arctan2.Open trajectory →
Question
The images are taken continuously from a first-person perspective. Based on your observation, choose the correct description from the options (it is known that the shooting time is during the spring equinox, in the morn…
Summary
Composes 3D reconstruction with mask segmentation, computes cross products.
Open trajectory →
Question
Measuring from the closest point of each object, what is the distance between the oven and the refridgerator (in meters)? Answer the spatial reasoning question about the 3D scene. For multiple-choice, answer with the o…
Summary
Composes 3D reconstruction with mask segmentation, reaches into
scipy.spatial.Open trajectory →
Question
When I sit at the round dining table facing the floor-to-ceiling window, I am facing south. In what direction are the kitchen and living room relative to me? Select the best answer from the given options. Answer with a…
Summary
Composes 3D reconstruction with mask segmentation, computes cross products. 85 cross-step variable references demonstrate state persistence across steps.
Open trajectory →
Question
In this video clip, how is the runner's location moving in the 3D scene relative to his/her/its starting orientation and location? This is a multiple-choice question about how an object or the camera moves in 3D space …
Summary
Composes 3D reconstruction with mask segmentation, computes cross products.
Open trajectory →
Question
Measuring from the closest point of each object, what is the distance between the chair and the radiator (in meters)? Answer the spatial reasoning question about the 3D scene. For multiple-choice, answer with the optio…
Summary
Composes 3D reconstruction with mask segmentation, reaches into
scipy.spatial.Open trajectory →
Question
Is there a moment in the video that shows a bright flash? Answer with a single letter (A, B, C, or D) corresponding to the correct choice. A. no B. yes
Summary
Multi-step trajectory (5 steps), plots intermediate state with
matplotlib.Open trajectory →
Question
What is the length of the longest dimension (length, width, or height) of the telephone, measured in centimeters? Answer the spatial reasoning question about the 3D scene. For multiple-choice, answer with the option's …
Summary
Composes 3D reconstruction with mask segmentation, reaches into
scipy.spatial.Open trajectory →
Question
In this video clip, how is the boat's location moving in the 3D scene relative to his/her/its starting orientation and location? This is a multiple-choice question about how an object or the camera moves in 3D space wi…
Summary
Composes 3D reconstruction with mask segmentation, computes cross products. 6 targeted VLM queries.
Open trajectory →
Question
During the sequence between frame 16 and frame 31 of 32, what was the primary consistent direction of the camera's movement relative to its orientation at the start? The options are Forward, Right, Backward, and Left. …
Summary
Builds a 3D reconstruction across frames, plots intermediate state with
matplotlib.Open trajectory →
Question
In this video clip, how is the character's location moving in the 3D scene relative to his/her/its starting orientation and location? This is a multiple-choice question about how an object or the camera moves in 3D spa…
Summary
Composes 3D reconstruction with mask segmentation, computes cross products.
Open trajectory →
Question
In this video clip, how is the race car's location moving in the 3D scene relative to his/her/its starting orientation and location? This is a multiple-choice question about how an object or the camera moves in 3D spac…
Summary
Composes 3D reconstruction with mask segmentation, computes cross products. 5 targeted VLM queries.
Open trajectory →
Question
What can be said about the camera movement in the video? Answer with a single letter (A, B, C, or D) corresponding to the correct choice. A. It zooms out B. It moves from bird's eye to eye level C. It tilts D. It zooms…
Summary
Builds a 3D reconstruction across frames, plots intermediate state with
matplotlib.Open trajectory →
Question
In this video clip, how is the man's location moving in the 3D scene relative to his/her/its starting orientation and location? This is a multiple-choice question about how an object or the camera moves in 3D space wit…
Summary
Composes 3D reconstruction with mask segmentation, computes cross products.
Open trajectory →
Question
Standing at television, gazing at toilet, where should desk be? A. front-right B. back-left C. front-left D. right This question is from ViewSpatial-Bench. The task tests spatial localization either from the camera's …
Summary
Composes 3D reconstruction with mask segmentation, computes cross products. 5 targeted VLM queries.
Open trajectory →
Question
In this video clip, how is the horse's location moving in the 3D scene relative to his/her/its starting orientation and location? This is a multiple-choice question about how an object or the camera moves in 3D space w…
Summary
Composes 3D reconstruction with mask segmentation, computes cross products.
Open trajectory →
Question
Seen from Car No.845, where is the man wearing the black hat? Answer with the option letter (A, B, C, or D) corresponding to the correct choice. A. right B. can not determine C. left D. can not determine
Summary
Composes 3D reconstruction with mask segmentation, computes cross products.
Open trajectory →
Question
Imagine standing at nightstand looking towards paper, where is clothes? A. right B. back C. front D. front-left This question is from ViewSpatial-Bench. The task tests spatial localization either from the camera's own…
Summary
Composes 3D reconstruction with mask segmentation, computes cross products. Iterative visual inspection: 10
show() calls. 5 targeted VLM queries.Open trajectory →
Question
When positioned at table facing window, where can you find whiteboard? A. back-left B. front-right C. front-left D. back-right This question is from ViewSpatial-Bench. The task tests spatial localization either from t…
Summary
Composes 3D reconstruction with mask segmentation, computes cross products.
Open trajectory →
Question
When positioned at sink facing counter, where can you find box? A. back-left B. back C. front-left D. right This question is from ViewSpatial-Bench. The task tests spatial localization either from the camera's own per…
Summary
Composes 3D reconstruction with mask segmentation, computes cross products.
Open trajectory →
Question
If a scene satisfies: ①There are two people in the scene, ②the woman is looking at the man, and ③a bright lamp is to the rear-left of the man. In what direction is the microphone relative to the man? Answer the single-…
Summary
Composes 3D reconstruction with mask segmentation, computes cross products.
Open trajectory →
Question
The video captures the actions of a person, during which objects are moved. How are these objects moved? (Considering the direction from the staircase to the camera as north) Select the best answer from the given optio…
Summary
Composes 3D reconstruction with mask segmentation, computes cross products. 11 targeted VLM queries.
Open trajectory →
Question
How does the relationship between trash bin (red bbox) and toilet (yellow bbox) evolve as the observer moves to toilet brush (green bbox) and orients towards toilet paper dispenser (blue bbox)? Calculate or judge based …
Summary
Composes 3D reconstruction with mask segmentation, computes cross products. 6 targeted VLM queries.
Open trajectory →
Question
<image> <image> How did the camera move to go from the first image to the second image? Choices: A. translate right. B. translate left. C. pan left. D. pan right. Please answer directly with only the letter of the corr…
Summary
Builds a 3D reconstruction across frames, computes angles with
arctan2.Open trajectory →
Question
Based on these four images (image 1, 2, 3, and 4) showing the green ball from different viewpoints (front, left, back, and right), with each camera aligned with room walls and partially capturing the surroundings: From …
Summary
Composes 3D reconstruction with mask segmentation. 5 targeted VLM queries.
Open trajectory →
Question
You are given several 2D points in an image, where (x, y) coordinates are normalized to [0, 1000]. A reference point is located at (457, 523). The following are four other points: - p0: (14, 204) - p1: (601, 75) - p2: …
Summary
Pairs 3D reconstruction with geometry primitives.
Open trajectory →
Question
At the beginning of the video, there are three little foxes: one stays still at the entrance of the den, while the other two are running. From the perspective of the stationary fox, in which direction are the other two …
Summary
Combines SAM3 segmentation with geometry primitives. 5 targeted VLM queries.
Open trajectory →
Question
How many green train carts are in the picture? Select the best answer from the given options. Answer with a single letter in parentheses, e.g. (A). A. 1 B. 0 C. 3 D. 2
Summary
Combines SAM3 segmentation with geometry primitives.
Open trajectory →
Question
In both videos, does the setting emphasize a vibrant and colorful environment with lively backgrounds? Yes. No. This sample contains 2-4 separate videos, each exposed in the kernel as its own `InputImages_<N>` (1-index…
Summary
Multi-step trajectory (5 steps).
Open trajectory →
Question
What is the raven doing in the video? Answer with a single letter (A, B, C, or D) corresponding to the correct choice. A. Eating. B. Flying. C. Walking. D. Sleeping.
Summary
Uses targeted segmentation.
Open trajectory →