SYNTOLOGY HomeExplorerAtlasCodeMethodologyAboutDevelopersFeedPricing
Paper · 2007.03639 · 2020

Human Trajectory Forecasting in Crowds: A Deep Learning Perspective

arXiv · PDF · Open in the Atlas

Code that ran

9 of 10 samples (functions and classes) lifted from this paper's linked repositories executed on a synthesized input in Syntology's jail (0 of them constructed an object rather than computing a result; 0 honoured the contract we drafted, 0 violated it and 9 have no contract verdict; 0 failures of our instrument; 1 no recorded run), measured 2026-11-22; repository links from the Papers with Code archive attributed under CC BY-SA, no affiliation; bodies inline only under a permissive licence as GitHub detects it. Not a reproduction of the paper's results.

"Ran" means the function executed on a synthesized input and returned a value. It is not a reproduction of the paper's results. This record, dated, as JSON: /api/ran/2007.03639.json · badge: /api/ran/2007.03639.svg

Each repository's role is a label: a classifier's output recorded on the link, or, where no role is recorded, what the Papers with Code archive marked. The line under it names its basis. Neither says who wrote the code.

RepositoryRoleRan
vita-epfl/trajnetplusplusbaselines community (classifier label)
classifier basis not recorded on the link
9 ran · 1 no recorded run
FunctionStatusWhere it lives
add_gt_observation_to_prediction Ran
ran on a synthesized input
vita-epfl/trajnetplusplusbaselines/evaluator/visualize_prediction_as_gif.py
code served (permissive licence) · get_code("f9747c27eac2f941")
How it ran

Ran in Syntology's sandbox on this synthesized input (drafted by a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API) and returned list, ndim 3, dtype family float; no output shape is stated. It ran in syntology-harvest-sandbox:v5-deps (sha256:6a292aa1aed1) at 2026-09-26T22:17:14Z. If your copy returns the same type and ndim on these inputs, it matched what we observed; nothing here checks that. It does NOT mean the output is correct: nothing has compared it to the paper.

import torch
import numpy as np

torch.manual_seed(0)

# Analyzing the function:
# - gt_observation is indexed by [ped_id] and [0], then sliced
# - model_prediction is indexed by [ped_id] and [0], then sliced
# - Both appear to be lists/sequences where each element can be indexed and sliced
# - obs_length = len(gt_observation[0]) - len(model_prediction[0])
# - The function concatenates gt_observation[ped_id][obs_length - 3:obs_length] with pred
# - enumerate(model_prediction) is used, so model_prediction must be iterable

# gt_observation[0] should have more elements than model_prediction[0]
# Let's say gt_observation[0] has 10 elements, model_prediction[0] has 6 elements
# obs_length = 10 - 6 = 4

# Each pedestrian's path is a list of coordinates
# gt_observation: list of pedestrian paths, each path is a list of coordinate tuples
# model_prediction: list of pedestrian paths, each path is a list of coordinate tuples

gt_observation = [
    [(0.0, 0.0), (1.0, 1.0), (2.0, 2.0), (3.0, 3.0), (4.0, 4.0), 
     (5.0, 5.0), (6.0, 6.0), (7.0, 7.0), (8.0, 8.0), (9.0, 9.0)],
    [(0.5, 0.5), (1.5, 1.5), (2.5, 2.5), (3.5, 3.5), (4.5, 4.5), 
     (5.5, 5.5), (6.5, 6.5), (7.5, 7.5), (8.5, 8.5), (9.5, 9.5)]
]

model_prediction = [
    [(10.0, 10.0), (11.0, 11.0), (12.0, 12.0), (13.0, 13.0), (14.0, 14.0), (15.0, 15.0)],
    [(10.5, 10.5), (11.5, 11.5), (12.5, 12.5), (13.5, 13.5), (14.5, 14.5), (15.5, 15.5)]
]

ARGS = {
    "gt_observation": gt_observation,
    "model_prediction": model_prediction
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
numpy, torch — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type list, ndim 3, dtype family float — a list output has no array shape of its own; none is stated
dtype family
int means an integer dtype; float means any other dtype the instrument saw, bool and complex included (fixture_check.py records only these two families)
Environment
image syntology-harvest-sandbox:v5-deps (sha256:6a292aa1aed1); not recorded for this sample: the harness installs a cuda-to-cpu shim (cuda_shim.py) before every sample and no run records whether a sample needed it; seed: none by the harness; any seed is in the setup block
ade Ran
ran on a synthesized input
vita-epfl/trajnetplusplusbaselines/evaluator/eval_utils.py
code served (permissive licence) · get_code("5ec1c642602bf41e")
How it ran

Ran in Syntology's sandbox on this synthesized input (drafted by a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API) and returned float32, ndim 0, dtype family float, shape () as a later run recorded it. The image it ran in was not recorded. If your copy returns the same type, ndim and shape on these inputs, it matched what we observed; nothing here checks that. It does NOT mean the output is correct: nothing has compared it to the paper.

import numpy as np

# The function expects:
# pred: Num_ped x Num_timesteps x 2
# gt: Num_ped x Num_timesteps x 2
# It accesses pred[0] and gt[0], which means it indexes the first pedestrian
# and expects 2D arrays of shape (Num_timesteps, 2)

# Create small but distinct-sized arrays
num_ped = 3
num_timesteps = 5

# pred: 3 pedestrians x 5 timesteps x 2 coordinates
pred = np.random.randn(num_ped, num_timesteps, 2).astype(np.float32)

# gt: 3 pedestrians x 5 timesteps x 2 coordinates
gt = np.random.randn(num_ped, num_timesteps, 2).astype(np.float32)

ARGS = {
    "pred": pred,
    "gt": gt,
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
numpy — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type float32, ndim 0, dtype family float, shape () — shape () as a later run recorded it
dtype family
int means an integer dtype; float means any other dtype the instrument saw, bool and complex included (fixture_check.py records only these two families)
Environment
image not recorded; re-verify against syntology-harvest-sandbox:v5-deps; not recorded for this sample: the harness installs a cuda-to-cpu shim (cuda_shim.py) before every sample and no run records whether a sample needed it; seed: none by the harness; any seed is in the setup block
collision Ran
ran on a synthesized input
vita-epfl/trajnetplusplusbaselines/evaluator/eval_utils.py
code served (permissive licence) · get_code("23245e535f55c843")
How it ran

Ran in Syntology's sandbox on this synthesized input (drafted by a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API) and returned bool, ndim 0, dtype family float, shape () as a later run recorded it. The image it ran in was not recorded. If your copy returns the same type, ndim and shape on these inputs, it matched what we observed; nothing here checks that. It does NOT mean the output is correct: nothing has compared it to the paper.

import numpy as np

# Create two simple non-colliding paths
# path1 and path2 should be Num_timesteps x 2 arrays
path1 = np.array([
    [0.0, 0.0],
    [1.0, 1.0],
    [2.0, 2.0],
    [3.0, 3.0]
])

path2 = np.array([
    [5.0, 5.0],
    [6.0, 6.0],
    [7.0, 7.0],
    [8.0, 8.0]
])

ARGS = {
    "path1": path1,
    "path2": path2,
    "person_radius": 0.1,
    "inter_parts": 2
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
numpy — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type bool, ndim 0, dtype family float, shape () — shape () as a later run recorded it
dtype family
int means an integer dtype; float means any other dtype the instrument saw, bool and complex included (fixture_check.py records only these two families)
Environment
image not recorded; re-verify against syntology-harvest-sandbox:v5-deps; not recorded for this sample: the harness installs a cuda-to-cpu shim (cuda_shim.py) before every sample and no run records whether a sample needed it; seed: none by the harness; any seed is in the setup block
fde Ran
ran on a synthesized input
vita-epfl/trajnetplusplusbaselines/evaluator/eval_utils.py
code served (permissive licence) · get_code("7f58d71fa664e242")
How it ran

Ran in Syntology's sandbox on this synthesized input (drafted by a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API) and returned float32, ndim 0, dtype family float, shape () as a later run recorded it. The image it ran in was not recorded. If your copy returns the same type, ndim and shape on these inputs, it matched what we observed; nothing here checks that. It does NOT mean the output is correct: nothing has compared it to the paper.

import numpy as np

# Construct valid arguments for the fde function
# The function expects:
# - pred: Num_ped x Num_timesteps x 2
# - gt: Num_ped x Num_timesteps x 2

# Create small sample arrays
# Using distinct sizes: 3 pedestrians, 5 timesteps, 2 coordinates
num_ped = 3
num_timesteps = 5

# Create prediction array (Num_ped x Num_timesteps x 2)
pred = np.random.randn(num_ped, num_timesteps, 2).astype(np.float32)

# Create ground truth array (Num_ped x Num_timesteps x 2)
gt = np.random.randn(num_ped, num_timesteps, 2).astype(np.float32)

ARGS = {
    "pred": pred,
    "gt": gt
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
numpy — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type float32, ndim 0, dtype family float, shape () — shape () as a later run recorded it
dtype family
int means an integer dtype; float means any other dtype the instrument saw, bool and complex included (fixture_check.py records only these two families)
Environment
image not recorded; re-verify against syntology-harvest-sandbox:v5-deps; not recorded for this sample: the harness installs a cuda-to-cpu shim (cuda_shim.py) before every sample and no run records whether a sample needed it; seed: none by the harness; any seed is in the setup block
get_dest Ran
ran on a synthesized input
vita-epfl/trajnetplusplusbaselines/get_dest.py
code served (permissive licence) · get_code("d07b9469e61a255f")
How it ran

Ran in Syntology's sandbox on this synthesized input (drafted by a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API) and returned dict. The image it ran in was not recorded. If your copy returns the same type on these inputs, it matched what we observed; nothing here checks that. It does NOT mean the output is correct: nothing has compared it to the paper.

import torch
import numpy as np

class MockRows:
    def __init__(self, data):
        self.data = data
    
    def collect(self):
        return self.data

# Create sample data: list of [frame, ped_id, x, y]
rows_data = [
    [0, 1, 1.0, 2.0],
    [1, 1, 1.5, 2.5],
    [2, 1, 2.0, 3.0],
    [0, 2, 5.0, 6.0],
    [1, 2, 5.5, 6.5],
    [2, 2, 6.0, 7.0],
]

rows = MockRows(rows_data)

ARGS = {
    'rows': rows
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
numpy, torch — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type dict — a dict output has no array shape of its own; none is stated — the instrument coerces any output to an array; for a dict output the recorded ndim 0 and dtype family float describe that coercion, not the output, and are not stated
Environment
image not recorded; re-verify against syntology-harvest-sandbox:v5-deps; not recorded for this sample: the harness installs a cuda-to-cpu shim (cuda_shim.py) before every sample and no run records whether a sample needed it; seed: none by the harness; any seed is in the setup block
preprocess_test Ran
ran on a synthesized input
vita-epfl/trajnetplusplusbaselines/evaluator/write_utils.py
code served (permissive licence) · get_code("4350795c816c54d7")
How it ran

Ran in Syntology's sandbox on this synthesized input (drafted by a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API) and returned list. The image it ran in was not recorded. If your copy returns the same type on these inputs, it matched what we observed; nothing here checks that. It does NOT mean the output is correct: nothing has compared it to the paper.

from collections import namedtuple

# Create a simple Row class that has a frame attribute
Row = namedtuple('Row', ['frame'])

# Create scene data structure
# scene is a list of pedestrian trajectories
# each trajectory is a list of Row objects with frame numbers
scene = [
    # First pedestrian: frames 0, 1, 2, 3, 4
    [Row(frame=0), Row(frame=1), Row(frame=2), Row(frame=3), Row(frame=4)],
    # Second pedestrian: frames 1, 2, 3, 4, 5
    [Row(frame=1), Row(frame=2), Row(frame=3), Row(frame=4), Row(frame=5)],
    # Third pedestrian: frames 2, 3, 4, 5, 6 (will be filtered if obs_len is small)
    [Row(frame=2), Row(frame=3), Row(frame=4), Row(frame=5), Row(frame=6)]
]

# obs_len is the observation length
obs_len = 3

ARGS = {
    'scene': scene,
    'obs_len': obs_len
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
none outside the standard library — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type list — a list output has no array shape of its own; none is stated
Environment
image not recorded; re-verify against syntology-harvest-sandbox:v5-deps; not recorded for this sample: the harness installs a cuda-to-cpu shim (cuda_shim.py) before every sample and no run records whether a sample needed it; seed: none by the harness; any seed is in the setup block
random_rotation Ran
ran on a synthesized input
vita-epfl/trajnetplusplusbaselines/trajnetbaselines/augmentation.py
code served (permissive licence) · get_code("a15d7a55c30696de")
How it ran

Ran in Syntology's sandbox on this synthesized input (drafted by a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API) and returned ndarray, ndim 3, dtype family float, shape (2, 4, 2) as a later run recorded it. The image it ran in was not recorded. If your copy returns the same type, ndim and shape on these inputs, it matched what we observed; nothing here checks that. It does NOT mean the output is correct: nothing has compared it to the paper.

import math
import random
import numpy

# Set seed for reproducibility
random.seed(0)
numpy.random.seed(0)

# Create a small xy array with shape (p, t, c)
# p = 2 (batch size), t = 4 (time steps), c = 2 (coordinates for 2D rotation)
xy = numpy.random.randn(2, 4, 2).astype(numpy.float32)

ARGS = {
    'xy': xy
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
numpy — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type ndarray, ndim 3, dtype family float, shape (2, 4, 2) — shape (2, 4, 2) as a later run recorded it
dtype family
int means an integer dtype; float means any other dtype the instrument saw, bool and complex included (fixture_check.py records only these two families)
Environment
image not recorded; re-verify against syntology-harvest-sandbox:v5-deps; not recorded for this sample: the harness installs a cuda-to-cpu shim (cuda_shim.py) before every sample and no run records whether a sample needed it; seed: none by the harness; any seed is in the setup block
trackrows Ran
ran on a synthesized input
vita-epfl/trajnetplusplusbaselines/get_dest.py
code served (permissive licence) · get_code("347bcda0c1310dd0")
How it ran

Ran in Syntology's sandbox on this synthesized input (drafted by a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API) and returned list, ndim 1, dtype family float; no output shape is stated. The image it ran in was not recorded. If your copy returns the same type and ndim on these inputs, it matched what we observed; nothing here checks that. It does NOT mean the output is correct: nothing has compared it to the paper.

import json

def trackrows(line):
    line = json.loads(line)
    track = line.get('track')
    if track is not None:
        return [track['f'], track['p'], track['x'], track['y']]
    return None

# Create a valid argument for the trackrows function
# The function expects a JSON string with an optional 'track' key
# containing a dict with 'f', 'p', 'x', 'y' keys

track_data = {
    'track': {
        'f': 1.5,
        'p': 2.3,
        'x': 10,
        'y': 20
    }
}

line_argument = json.dumps(track_data)

ARGS = {
    'line': line_argument
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
none outside the standard library — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type list, ndim 1, dtype family float — a list output has no array shape of its own; none is stated
dtype family
int means an integer dtype; float means any other dtype the instrument saw, bool and complex included (fixture_check.py records only these two families)
Environment
image not recorded; re-verify against syntology-harvest-sandbox:v5-deps; not recorded for this sample: the harness installs a cuda-to-cpu shim (cuda_shim.py) before every sample and no run records whether a sample needed it; seed: none by the harness; any seed is in the setup block
update_lines Ran
ran on a synthesized input
vita-epfl/trajnetplusplusbaselines/evaluator/visualize_prediction_as_gif.py
code served (permissive licence) · get_code("cd91bf4e42e6a862")
How it ran

Ran in Syntology's sandbox on this synthesized input (drafted by a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API) and returned list, ndim 1, dtype family float; no output shape is stated. The image it ran in was not recorded. If your copy returns the same type and ndim on these inputs, it matched what we observed; nothing here checks that. It does NOT mean the output is correct: nothing has compared it to the paper.

import torch
import numpy as np
from matplotlib.lines import Line2D

torch.manual_seed(0)

# Create mock Line2D objects that have set_data and set_3d_properties methods
class MockLine:
    def __init__(self):
        self.xdata = None
        self.ydata = None
        self.zdata = None
    
    def set_data(self, data):
        # data should be shape (2, num) - x and y coordinates
        self.xdata = data[0]
        self.ydata = data[1]
    
    def set_3d_properties(self, data):
        # data should be 1D array of z coordinates
        self.zdata = data

# Create sample data
# dataLines should be a list/array where each element is shape (3, num_points)
# representing [x, y, z] coordinates for each point
num_points = 5
data1 = np.random.randn(3, num_points)
data2 = np.random.randn(3, num_points)
dataLines = [data1, data2]

# Create line objects
line1 = MockLine()
line2 = MockLine()
lines = [line1, line2]

# num parameter - integer specifying how many points to plot
num = 3

ARGS = {
    "num": num,
    "dataLines": dataLines,
    "lines": lines
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
matplotlib, numpy, torch — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type list, ndim 1, dtype family float — a list output has no array shape of its own; none is stated
dtype family
int means an integer dtype; float means any other dtype the instrument saw, bool and complex included (fixture_check.py records only these two families)
Environment
image not recorded; re-verify against syntology-harvest-sandbox:v5-deps; not recorded for this sample: the harness installs a cuda-to-cpu shim (cuda_shim.py) before every sample and no run records whether a sample needed it; seed: none by the harness; any seed is in the setup block
add_gt_observation_to_prediction Not yet run
unverified (harvested, no recorded run)
vita-epfl/trajnetplusplusbaselines/evaluator/visualize_predictions.py
code served (permissive licence) · get_code("df36c6556782b4b9")
How it ran

no invocation recorded.

Also present in other papers' code

Of 10 samples lifted from this paper's linked repositories, 10 carry a function identity and 0 do not. Of the 10 distinct functions with an identity, 0 are also present in other papers whose code we hold (0 with at least one paper dated earlier, 0 the same month, 0 later) and 10 are only here among the 52,977 papers whose code Syntology holds (as of 2026-09-26).

Only here means: among the 52,977 papers whose code Syntology holds (51,887 with lifted samples, 1,090 through their assemblies only), as of 2026-09-26, no other paper's lifted samples or current assembly bricks contain an identical function body after normalisation.

Presence, not direction: an identical function body in two papers' repositories does not say who wrote it, which came from which, or that either paper's method rests on it. Matching is exact after normalisation, so a renamed or edited function is not matched. Dates are when papers appeared, not when code was written. Only .py files the harvester lifted are covered; notebooks are not read. A new-style arXiv id YYMM.NNNN dates a paper to the month 20YY-MM; otherwise Paper.year dates it to the year; otherwise it is undated. Dates are when the paper appeared, not when the code was written. Read 2026-11-22.

Repositories linked to this paper

Some links come from the archived Papers with Code dataset (CC BY-SA 4.0): attribution and licence.

Abstract

Since the past few decades, human trajectory forecasting has been a field of active research owing to its numerous real-world applications: evacuation situation analysis, deployment of intelligent transport systems, traffic operations, to name a few. Early works handcrafted this representation based on domain knowledge. However, social interactions in crowded environments are not only diverse but often subtle. Recently, deep learning methods have outperformed their handcrafted counterparts, as they learned about human-human interactions in a more generic data-driven fashion. In this work, we present an in-depth analysis of existing deep learning-based methods for modelling social interactions. We propose two knowledge-based data-driven methods to effectively capture these social interactions. To objectively compare the performance of these interaction-based forecasting models, we develop a large scale interaction-centric benchmark TrajNet++, a significant yet missing component in the field of human trajectory forecasting. We propose novel performance metrics that evaluate the ability of a model to output socially acceptable trajectories. Experiments on TrajNet++ validate the need for our proposed metrics, and our method outperforms competitive baselines on both real-world and synthetic datasets.

For agents

The same record, over MCP at https://syntology.ai/mcp:

get_harvested_code_for_paper("2007.03639")
get_code_for_paper("2007.03639")
have("2007.03639")

The run record, dated, one paper per request, free:

curl https://syntology.ai/api/ran/2007.03639.json

A badge for a README (the split and the date, never a ratio):

[![Syntology run record](https://syntology.ai/api/ran/2007.03639.svg)](https://syntology.ai/paper/2007.03639)

Connect an agent — have() is free.