Update README.md

README.md

@@ -24,37 +24,6 @@ tags:
 - tabular
 - timeseries
 - image
-size_categories:
-- 1B<n<10B
----
-U-EASWS Dataset Card
----
----
-pretty_name: >-
-  U-EASWS: An Urban Expressway Type-A Short Weaving Segment UAV Vehicle
-  Trajectory Dataset Integrated with Interaction and Risk Semantic Annotations
-language:
-  - en
-license: cc-by-4.0
-source_datasets:
-  - original
-tags:
-  - vehicle-trajectory
-  - uav
-  - urban-expressway
-  - weaving-segment
-  - type-a-weaving
-  - short-weaving-segment
-  - microscopic-traffic
-  - lane-changing
-  - traffic-safety
-  - surrogate-safety-measures
-  - vehicle-interaction
-  - interaction-semantics
-  - risk-semantics
-  - tabular
-  - timeseries
-  - image
 ---
 
 # U-EASWS
@@ -63,7 +32,7 @@ tags:
 
 U-EASWS (Urban Expressway Type-A Short Weaving Segment UAV Vehicle Trajectory Dataset Integrated with Interaction and Risk Semantic Annotations) is a UAV-based microscopic vehicle trajectory dataset specifically developed for urban expressway Type-A short weaving segments.
 
-The dataset contains 19,122 high-quality vehicle trajectories collected from seven real-world Type-A short weaving scenarios in Changchun, China, with approximately 3.6 h of valid aerial video observations. The observed weaving segments cover different roadway geometries and channelization configurations, including unrestricted weaving, restricted merging, and restricted diverging layouts.
+The dataset contains 19,122 vehicle trajectories collected from seven real-world Type-A short weaving scenarios in Changchun, China, with approximately 3.6 h of segment-level observation coverage when the seven scenario durations are summed. This sum counts the two directions of each shared bidirectional recording separately; it is not the duration of unique video footage. The observed weaving segments cover different roadway geometries and channelization configurations, including unrestricted weaving, restricted merging, and restricted diverging layouts.
 
 Type-A weaving is characterized by merging and diverging traffic streams that are each required to perform a lane change within the same weaving area. In short weaving segments, the limited longitudinal distance available for completing these maneuvers produces frequent mandatory lane-changing interactions among merging, diverging, and through vehicles. U-EASWS was therefore designed specifically to provide continuous microscopic observations of vehicle movements under these geometrically constrained and interaction-intensive conditions.
 
@@ -89,13 +58,18 @@ U-EASWS is intended to support research on microscopic traffic flow and driving
 
 ## Dataset Repository Structure
 
-The repository is organized according to the processing level and semantic content of the data. The current release contains five main data components.
+The repository is organized according to the processing level and semantic content of the data. The current release contains five data/documentation components and a principal processing-code folder.
 
 ### 1. Weaving Segment Geometry Images
 
 [Weaving_Segment_Geometry_Images](https://huggingface.co/datasets/InterestingITS/U-EASWS/tree/main/Weaving_Segment_Geometry_Images)
 
-This folder contains seven top-down aerial channelization images corresponding to Weaving Segments WS1–WS7.
+This folder contains four top-down aerial geometry images covering seven weaving segments:
+
+- [WS1_2.png](https://huggingface.co/datasets/InterestingITS/U-EASWS/resolve/main/Weaving_Segment_Geometry_Images/WS1_2.png): WS1 and WS2.
+- [WS3.png](https://huggingface.co/datasets/InterestingITS/U-EASWS/resolve/main/Weaving_Segment_Geometry_Images/WS3.png): WS3.
+- [WS4_5.png](https://huggingface.co/datasets/InterestingITS/U-EASWS/resolve/main/Weaving_Segment_Geometry_Images/WS4_5.png): WS4 and WS5.
+- [WS6_7.png](https://huggingface.co/datasets/InterestingITS/U-EASWS/resolve/main/Weaving_Segment_Geometry_Images/WS6_7.png): WS6 and WS7.
 
 The images provide the roadway geometry and channelization context required to interpret the trajectory data. They show the actual road layout, lane arrangement, lane boundaries, solid and dashed lane markings, and the geometric organization of the observed Type-A short weaving segments.
 
@@ -128,7 +102,7 @@ The raw data are divided into 16 CSV files:
 - `WS6_7_raw_detection_data_2.csv`
 - `WS6_7_raw_detection_data_3.csv`
 
-Files beginning with `WS1_2`, `WS4_5`, and `WS6_7` correspond to the three bidirectional observation sites, whereas files beginning with `WS3` correspond to the one-way observation scenario.
+Files beginning with `WS1_2`, `WS4_5`, and `WS6_7` correspond to the three bidirectional observation sites, whereas files beginning with `WS3` correspond to the scenario retaining one traffic direction.
 
 Each row represents one tracked vehicle observation in a particular video frame.
 
@@ -143,7 +117,7 @@ The released raw data contain the following fields:
 
 The raw files preserve image-space vehicle positions and detection-box dimensions and can therefore be used for independent coordinate transformation, vehicle-dimension estimation, trajectory interpolation, smoothing, or other user-defined post-processing procedures.
 
-The `Vehicle_ID` in the raw data is the tracking identifier associated with the corresponding source sequence and should be distinguished from the globally unique `Ego_Vehicle_ID` provided in the processed trajectory files.
+The `Vehicle_ID` in the raw data is the tracking identifier associated with the corresponding source sequence and should be distinguished from the `Ego_Vehicle_ID` provided in the processed trajectory files. Vehicle identifiers are not unique across observation-group files; retain the file/source identity when combining groups.
 
 The term "raw data" in this repository specifically refers to frame-level vehicle detection and tracking outputs. It does not refer to the original UAV video files.
 
@@ -162,9 +136,9 @@ The files provide reconstructed vehicle trajectories expressed in a road-aligned
 
 The main fields are defined as follows:
 
-- `Ego_Vehicle_ID`: globally unique vehicle identifier. Unit: dimensionless identifier.
-- `Frame`: source video frame index. Unit: dimensionless frame index.
-- `Delta_t_s`: physical time interval associated with the current trajectory record. Unit: s.
+- `Ego_Vehicle_ID`: processed vehicle identifier within the corresponding observation group. The same numerical ID can occur in another group. Unit: dimensionless identifier.
+- `Frame`: frame index in the corresponding processed sequence; source-to-processed mappings must be checked before linking to raw records. Unit: dimensionless frame index.
+- `Delta_t_s`: elapsed time from the previous record of the same vehicle, after ordering by frame; the first record has no preceding interval and is blank. Unit: s.
 - `x_center`: vehicle-center coordinate along the longitudinal road axis in the road-aligned local physical coordinate system. Unit: m.
 - `y_center`: vehicle-center coordinate along the lateral road axis in the road-aligned local physical coordinate system. Unit: m.
 - `v_x`: longitudinal vehicle velocity. Unit: m/s.
@@ -177,9 +151,9 @@ The main fields are defined as follows:
 - `Dist_to_Diverge_m`: absolute longitudinal distance from the current vehicle center to the predefined diverge reference cross-section. Unit: m.
 - `Vehicle_Length_m`: estimated physical length of the vehicle. Unit: m.
 
-`Vehicle_Length_m` is estimated as a vehicle-level physical attribute from multiple valid frame-level bounding-box observations and remains constant for the complete trajectory of the same uniquely identified vehicle.
+`Vehicle_Length_m` is estimated as a vehicle-level physical attribute from multiple valid frame-level bounding-box observations and remains constant for the complete trajectory of the same vehicle within an observation group.
 
-`Dist_to_Merge_m` and `Dist_to_Diverge_m` explicitly connect individual vehicle states to the principal geometric locations of the weaving segment and can be used to analyze how microscopic behavior changes with the remaining distance to the merge or diverge reference cross-section.
+`Dist_to_Merge_m` and `Dist_to_Diverge_m` explicitly connect individual vehicle states to the principal geometric locations of the weaving segment and can be used to analyze how microscopic behavior changes with the absolute distance to the merge or diverge reference cross-section. 
 
 ### 4. Risk Semantic Data
 
@@ -199,7 +173,7 @@ The additional fields are:
 - `TTC`: Time-to-Collision. Unit: s.
 - `PET`: Post-Encroachment Time. Unit: s.
 - `DRAC`: Deceleration Rate to Avoid a Crash. Unit: m/s².
-- `Max_Risk_Target_ID`: ID of the target vehicle producing the highest conflict risk for the ego vehicle in the current frame. The value is set to `-1` when no corresponding risk target exists. Unit: dimensionless identifier.
+- `Max_Risk_Target_ID`: ID of the target assigned by the risk-annotation rule in the current frame. The released code prioritizes a high-risk PET event target and otherwise uses the relevant leader/event target; this legacy field name does not denote a global severity ranking over all surrounding vehicles. The value is set to `-1` when no corresponding risk target exists. Unit: dimensionless identifier.
 - `Risk_Level`: explicit risk classification, with values `High Risk` or `Low Risk`.
 
 TTC and DRAC are frame-wise state-based surrogate safety measures. PET is treated as an event-level measure associated with identified cut-in events. For each identified cut-in event, the calculated PET value is assigned at the event-detection frame to both the cut-in vehicle and its corresponding following vehicle.
@@ -233,14 +207,20 @@ The additional fields are:
 - `x_local`: relative-position component of the target vehicle along the local X-axis of the vehicle-pair coordinate system. Unit: m.
 - `y_local`: relative-position component of the target vehicle along the local Y-axis of the vehicle-pair coordinate system. Unit: m.
 - `v_rel`: scalar relative speed between the vehicle pair. Unit: m/s.
-- `Interaction Strength`: dimensionless PR-based measure of proximity-induced vehicle interaction, ranging from 0 to 1. Larger values represent stronger proximity-induced interaction.
-- `Is Interacting`: binary indicator of whether the candidate vehicle pair satisfies the effective-interaction criterion. `1` denotes an effective interaction and `0` otherwise.
-- `Interaction Type`: behavioral interaction category, including `Car Following`, `Parallel Driving`, and `Lane Changing/Weaving`.
+- `Interaction Strength` (legacy column name `Interaction Intensity` in the WS3 and WS6–7 files): dimensionless PR-based measure of proximity-induced vehicle interaction, ranging from 0 to 1. Larger values represent stronger proximity-induced interaction.
+- `Is Interacting`: binary indicator of whether the candidate vehicle pair satisfies the effective-interaction criterion. `1` denotes PR greater than `exp(-1)` and `0` otherwise. Candidate pairs below this threshold remain in the annotated data.
+- `Interaction Type`: behavioral interaction category, including `Car Following`, `Parallel Driving`, and `Lane Changing/Weaving`. The released code uses the legacy value `Safe` for records without a supplied candidate; this is not a guarantee of traffic safety.
 
 `Interaction Strength` represents proximity-induced interaction rather than collision probability or conflict severity. A large PR value indicates that the target vehicle lies deeper within the empirically inferred driver space of the ego vehicle and therefore represents stronger spatial interaction.
 
 Consequently, `Is Interacting = 1` should not automatically be interpreted as a high-risk traffic conflict. Risk semantics and interaction semantics describe different and complementary dimensions of microscopic traffic behavior.
 
+### 6. Principal Processing Code
+
+[Principal Processing Code](https://huggingface.co/datasets/InterestingITS/U-EASWS/tree/main/Principal%20Processing%20Code)
+
+This folder contains the 11 principal processing and supporting files described in the reprocessing guide below. The code README provides detailed inputs, parameter settings, and implementation limits.
+
 ## Road Geometry and Scenario Information
 
 The dataset covers seven Type-A short weaving segments located on urban expressways in Changchun, China.
@@ -253,7 +233,7 @@ Three UAV observation sequences contain bidirectional traffic flows and each cor
 - WS4 and WS5;
 - WS6 and WS7.
 
-The remaining observation sequence contains one-way traffic and corresponds to WS3.
+The remaining observation group corresponds to WS3 and retains one traffic direction after preprocessing.
 
 WS1, WS3, WS4, and WS6 retain left-to-right vehicle movement in the road-geometry representation, whereas WS2, WS5, and WS7 represent the opposite right-to-left traffic direction.
 
@@ -340,7 +320,9 @@ Because U-EASWS contains observations originating from both frame rates, users s
 
 The processed trajectory files therefore include `Delta_t_s`, which explicitly provides the physical time interval associated with the corresponding trajectory record.
 
-`Frame` and `Delta_t_s` should be used together when reconstructing physical time sequences or calculating time-dependent variables.
+Within a vehicle trajectory, sort by `Frame` and accumulate valid `Delta_t_s` values to obtain elapsed time relative to the first observation. A blank first interval is expected. For interaction files, first remove repeated ego/frame observations introduced by multiple candidate targets; summing all pair rows would overcount time.
+
+The six-column raw CSV files do not include frame rate or timestamps. Their frame indices must be interpreted using source-specific recording metadata. WS3 contains mixed recording rates, so users should not assign a rate to an individual WS3 raw file solely from its numerical filename suffix. A verified per-source frame-rate mapping is required for raw-data reprocessing; it is not supplied by the calibration template.
 
 ## Data Collection
 
@@ -350,9 +332,9 @@ The recording modes were 3840 × 2160 at 30 fps and 2688 × 1512 at 60 fps. The
 
 The zoom function was not used during data collection in order to avoid changes in image scale and additional geometric distortion caused by zoom adjustment.
 
-The UAV recording altitude ranged approximately from 220 m to 360 m, while the retained observation lengths ranged approximately from 290 m to 400 m.
+The UAV recording altitude ranged approximately from 220 m to 360 m, while the retained observation lengths ranged approximately from 290 m to 375 m.
 
-Data collection was performed during daytime periods under clear and low-wind conditions to reduce severe vehicle-shadow effects and UAV-view instability.
+Data collection was performed during daytime periods under clear and low-wind conditions. Vehicle shadows and residual UAV-view instability can nevertheless affect image measurements.
 
 ## Data Processing
 
@@ -364,6 +346,127 @@ Vehicle trajectories are reconstructed using PCHIP interpolation where sufficien
 
 The processed data should therefore be distinguished from the frame-level records provided in `Raw_Data`.
 
+## Principal Processing Code and Reprocessing Guide
+
+The [Principal Processing Code folder](https://huggingface.co/datasets/InterestingITS/U-EASWS/tree/main/Principal%20Processing%20Code) contains 11 files. These are executable Python processing modules, supporting utilities, a calibration template, dependency requirements, and documentation. 
+
+### Files and Functions
+
+- **`coordinate_transform.py`** registers source-image coordinates to a master frame using a planar homography and converts pixel coordinates into scene-level metric coordinates. It requires a completed calibration JSON and the corresponding source identifier. It retains source identity and adds metric vehicle-center coordinates.
+- **`filter_trajectory_outliers.py`** screens metric trajectories using observed duration, observation count, spatial extent, and average path speed. It produces retained records, rejected records, and a per-vehicle screening summary. The same screening is already included in reconstruction, so separate execution is optional.
+- **`reconstruct_trajectories.py`** screens trajectories, fills missing frames, interpolates positions, fills lane/direction attributes, applies symmetric exponential moving average (sEMA) smoothing, and calculates velocities and accelerations. It also preserves source/original identifiers and produces timing information. Measured merge and diverge reference coordinates can optionally be supplied to generate the corresponding distance fields.
+- **`estimate_vehicle_length.py`** estimates a fixed physical length for each vehicle directly from the six-column raw detection records and the calibration JSON. It returns vehicle-level length estimates, the number of retained observations, and quality indicators.
+- **`merge_vehicle_lengths.py`** joins vehicle-level length estimates to reconstructed trajectories using source identity and original vehicle identity. A verified mapping file is required when those lineage fields are absent from the trajectory input.
+- **`annotate_risk.py`** calculates nearest-leader TTC and DRAC, event-level cut-in PET, target identifiers, and risk labels using physical trajectories, time information, lane/direction annotations, and vehicle lengths.
+- **`annotate_interactions.py`** calculates PR-based interaction strength, the effective-interaction flag, and interaction type from basic trajectories, precomputed candidate-pair states, and calibrated PR parameters. It retains supplied candidate pairs even when their PR does not satisfy the effective-interaction threshold.
+- **`_common.py`** provides shared CSV validation and output-writing utilities. Keep it alongside the processing scripts; it is imported by them and is not a standalone processing step.
+- **`calibration_template.json`** specifies the required calibration structure without providing actual scene measurements. Users must complete it before running coordinate transformation or vehicle-length estimation.
+- **`requirements.txt`** lists the processing dependencies and their versions.
+- **`README.md`** documents detailed input schemas, commands, parameter choices, and implementation limitations for the processing package. It is separate from this repository-level Dataset Card.
+
+### Installation
+
+Download the 11 files together, retaining their published filenames and their common directory. Python 3.12 was used for package testing. From the local repository directory, run:
+
+```bash
+cd "Principal Processing Code"
+python -m pip install -r requirements.txt
+python reconstruct_trajectories.py --help
+```
+
+Each processing script provides `--help`. The commands below use descriptive placeholders; replace them with actual local input/output filenames, the source key defined in your calibration JSON, and the numerical source frame rate. Output filenames must be new because the scripts do not overwrite existing files.
+
+### Inputs to Prepare
+
+1. **Raw detections:** one source recording with `Frame, Vehicle_ID, x_center_px, y_center_px, width_px, height_px`. Pixel coordinates must refer to the same image view used for calibration.
+2. **Scene calibration:** a completed copy of `calibration_template.json`, including the master source, master-frame origin, reference segments with measured physical distances, and corresponding control points for non-master sources. The template itself contains no usable site calibration.
+3. **Lane and direction annotations:** verified `Lane number` and `Direction` fields for the metric trajectories. The six raw columns do not determine lane assignment or road geometry.
+4. **Recording metadata:** the actual constant frame rate for each source, and explicit source/ID/time mappings when combining recordings. Optional merge/diverge distances require measured reference cross-sections in the same metric coordinate system.
+5. **Interaction inputs:** a candidate-pair CSV and calibrated PR parameter CSV, with schemas described below. Pair generation and PR parameter fitting are upstream steps not included in the released scripts.
+
+For calibration, replace the required `null` values with measured values and change `template_only` to `false` only after completing the configuration. At least four valid corresponding points are required for each non-master homography. The current implementation requires an already horizontally aligned road and `rotation_rad = 0`; it does not implement curved-road Frenet projection. The master-frame scale is the mean of the measured physical-to-pixel distance ratios. Metric x increases to the right of the aligned master image, while metric y reverses the image's downward axis.
+
+### Processing Commands
+
+First transform the pixel centers to meters:
+
+```bash
+python coordinate_transform.py --input RAW_DETECTION_DATA.csv --calibration SCENE_CALIBRATION.json --source SOURCE_VIDEO_KEY --output METRIC_TRAJECTORIES.csv
+```
+
+Add verified lane and direction annotations to that output, then reconstruct the trajectories:
+
+```bash
+python reconstruct_trajectories.py --input METRIC_TRAJECTORIES_WITH_LANES.csv --fps SOURCE_FPS --source SOURCE_VIDEO_KEY --output RECONSTRUCTED_TRAJECTORIES.csv --rejected REJECTED_TRAJECTORIES.csv --audit SCREENING_SUMMARY.csv
+```
+
+Estimate lengths from the original six-column detections using the same calibration, and merge them into the retained trajectories:
+
+```bash
+python estimate_vehicle_length.py --input RAW_DETECTION_DATA.csv --calibration SCENE_CALIBRATION.json --source SOURCE_VIDEO_KEY --output VEHICLE_LENGTH_ESTIMATES.csv
+python merge_vehicle_lengths.py --input RECONSTRUCTED_TRAJECTORIES.csv --lengths VEHICLE_LENGTH_ESTIMATES.csv --output BASIC_TRAJECTORY_DATA.csv
+```
+
+Generate the two semantic layers independently:
+
+```bash
+python annotate_risk.py --input BASIC_TRAJECTORY_DATA.csv --output RISK_SEMANTIC_DATA.csv
+python annotate_interactions.py --input BASIC_TRAJECTORY_DATA.csv --pairs CANDIDATE_PAIRS.csv --parameters PR_PARAMETERS.csv --output INTERACTION_SEMANTIC_DATA.csv
+```
+
+Process each constant-frame-rate recording separately. Do not treat independent recordings, or WS3 recordings with different frame rates, as one continuous video. When joining recordings, retain source identity and explicitly manage vehicle identifiers and physical time. The code does not infer historical mappings between raw tracking IDs and the identifiers in the published processed files.
+
+### Screening and Reconstruction Settings
+
+The released reconstruction code rejects a vehicle track when its observed duration `N/FPS` is below 0.5 s or it contains fewer than 15 valid observations. It also rejects a suspected low-motion track when all three conditions hold: `N/FPS >= 2 s`, the diagonal of its coordinate extent is below 1 m, and its average path speed is below 0.5 m/s. Here, N counts observations before interpolation; duration is not the elapsed time between the first and last frame. Average path speed is the sum of consecutive observed travel distances divided by `N/FPS`.
+
+These rules identify short tracks and suspected persistent low-motion detections. They do not explicitly measure movement irregularity and cannot reliably distinguish every false detection from a genuinely stopped vehicle. Inspect the rejected records and screening summary before accepting the filtering results.
+
+For retained tracks, exactly two valid coordinate observations select linear interpolation and three or more select PCHIP. With the default minimum-observation screen, two-point whole tracks are normally rejected before this interpolation branch is reached. Lane and direction values use forward fill followed by backward fill; vehicle identity remains constant within each trajectory. Position smoothing uses sEMA with a default time parameter of 0.6 s, followed by positional differences for velocity and acceleration.
+
+Newly reconstructed outputs retain `Source_ID`, `Original_Vehicle_ID`, and `Original_Frame`, together with `Time_s`. Their `Delta_t_s` is missing at the first record of each track and equals `1/FPS` afterward. These additional lineage/time columns may not be present in the published processed CSV schema.
+
+### Vehicle-Length Estimation Settings
+
+The estimator constructs bounding-box corners internally from the center coordinates and half the width/height. It transforms all four corners to metric coordinates and uses the longitudinal maximum-minus-minimum extent as a frame-level length candidate. This is a geometric calculation using the supplied boxes, not another detection or matching step.
+
+Nonfinite measurements and nonpositive box dimensions are excluded. With at least four valid candidates, a 1.5-IQR filter is applied. The 25th percentile of the retained candidates is assigned as the fixed vehicle length if at least five remain; otherwise the length is missing and its quality is marked `insufficient`. The output lateral extent is a bounding-box diagnostic. Box orientation and imperfect localization can affect these estimates; six pixel columns alone cannot determine physical length without metric calibration.
+
+Semantic-Annotation Inputs and Interpretation
+
+Risk annotation requires physical time, lane/direction information, and a valid fixed length for each vehicle. The current script uses the nearest same-lane leader for TTC/DRAC and trajectory crossings associated with cut-in events for PET. PET uses future observations and is an event-based retrospective measure. The default DRAC cap is 15 m/s². Missing vehicle lengths raise an error unless an explicit fallback is requested. Infinite TTC or missing PET should be interpreted according to the metric's applicability, not automatically as observed safety.
+
+Candidate-pair input requires Frame, EV_Vehicle_ID, POV_Vehicle_ID, x_local, y_local, v_rel, cangle. Pairs must refer to valid vehicles in the same source/frame, travel in the same direction, lie within 50 m, and have positive scalar relative speed. The interaction angle described in the manuscript is a continuous quantity. For the processing implementation, angles from 5° to 175°, inclusive, are encoded as cangle = 1, and angles outside this interval as cangle = 0. This conversion must be completed before supplying the candidate-pair file to the annotation script. The value 1 selects Lane Changing/Weaving; the value 0 selects Car Following for vehicles in the same lane and Parallel Driving otherwise.
+
+PR parameter input requires cangle, round_v, rx_plus, rx_minus, ry_plus, ry_minus, bx_plus, bx_minus, by_plus, by_minus. Scale and shape parameters must be positive. The script rounds relative speed to one decimal place and selects the nearest parameter bin within the matching cangle group. Candidate-pair inputs and PR parameter files must use the same angle encoding, local-coordinate convention, and calibration procedure.
+
+All supplied candidate pairs receive annotations. Interaction Strength > exp(-1) determines Is Interacting = 1; the threshold does not remove weaker candidate pairs. Rows without a supplied candidate use Interaction_Target_ID = -1, strength 0, flag 0, and the legacy type Safe. This type means no candidate was supplied for annotation and does not establish absence of traffic risk.
+
+### Public-Code Scope
+
+The package provides executable implementations of the principal post-processing stages. Original UAV videos, video stabilization, detector training and tracking, lane-region creation, candidate-pair generation/local projection, and maximum-likelihood PR fitting are not included. The repository provides a calibration template rather than actual per-source calibration measurements; the candidate-pair and calibrated PR parameter files needed for reprocessing are also not included in this release.
+
+Users can analyze the published processed data directly. Reprocessing the six-column raw records requires the additional scene/source inputs listed above. Outputs obtained using new calibration, filtering, mappings, or parameter choices need not reproduce the published values exactly. See the [processing README](https://huggingface.co/datasets/InterestingITS/U-EASWS/blob/main/Principal%20Processing%20Code/README.md) for detailed implementation behavior.
+
+## Reading and Selecting Data
+
+Download the CSV for the observation group and processing layer needed for your analysis. The interaction files contain multiple candidate-pair rows for some ego-vehicle/frame combinations and are substantially larger than the basic trajectory files. They should not be treated as one row per vehicle per frame.
+
+For example, after downloading a basic trajectory file:
+
+```python
+import pandas as pd
+
+filename = "Basic_Trajectory_Data/Weaving Segment_3.csv"
+for chunk in pd.read_csv(filename, chunksize=100_000):
+    print(chunk.head())
+    break
+```
+
+Use file/source identity together with vehicle and frame identifiers when linking data. Do not assume raw `Vehicle_ID` equals processed `Ego_Vehicle_ID`. For interaction rows, also retain `Interaction_Target_ID`; deduplicate ego/frame records before counting trajectories or reconstructing a vehicle's time axis. The geometry images are contextual documentation, not detection ground-truth annotations.
+
+The repository does not prescribe a common train/validation/test split across these trajectory layers. For predictive modeling, split by recording, vehicle, or scenario as appropriate to avoid leakage between neighboring frames and overlapping vehicle trajectories. Files from different processing layers have different columns and should be loaded separately.
+
 ## Relationship Between Data Components
 
 The repository provides multiple data levels so that users can select the level most appropriate for their research objective.
@@ -372,7 +475,7 @@ The repository provides multiple data levels so that users can select the level
 
 `Basic_Trajectory_Data` provides reconstructed physical vehicle trajectories and is suitable for vehicle kinematics, lane-changing behavior, traffic-flow modeling, trajectory prediction, and simulation calibration.
 
-`Risk_Semantic_Data` retains the basic trajectory information and supplements it with TTC, PET, DRAC, maximum-risk-target information, and explicit risk labels.
+`Risk_Semantic_Data` retains the basic trajectory information and supplements it with TTC, PET, DRAC, selected risk-target information, and explicit risk labels.
 
 `Interaction_Semantic_Data` retains the basic trajectory information and supplements it with vehicle-pair relative states, PR-based interaction strength, effective-interaction labels, and interaction-type annotations.
 
@@ -382,13 +485,13 @@ The processed risk and interaction files already retain the corresponding basic
 
 ## Technical Validation
 
-The quality of U-EASWS was evaluated from multiple complementary perspectives.
+The accompanying manuscript reports evaluation of U-EASWS from multiple complementary perspectives.
 
 Vehicle detection and multi-object tracking performance were examined under dense UAV-based traffic observations.
 
 Trajectory continuity and spatial consistency were evaluated after coordinate registration, reconstruction, and smoothing.
 
-Kinematic consistency was assessed using reconstructed position, velocity, and acceleration profiles to verify that the processed trajectories remain compatible with realistic microscopic vehicle motion.
+Kinematic consistency was assessed using reconstructed position, velocity, and acceleration profiles. Numerical differentiation can amplify positional noise, particularly at trajectory boundaries. Users should inspect derivative outliers before applying acceleration thresholds or estimating vehicle dynamics; the availability of validation analyses does not establish that every released record is free of such outliers.
 
 The spatial distributions of lane-changing events were further examined across all seven Type-A short weaving segments. The results show that lane-changing behavior is systematically related to roadway geometry, lane position, travel direction, and channelization constraints rather than being randomly distributed within the observation area.
 
@@ -434,13 +537,11 @@ The dataset covers seven Type-A short weaving scenarios in Changchun, China. Res
 
 U-EASWS is licensed under the Creative Commons Attribution 4.0 International License (CC BY 4.0).
 
-Under this license, users may share and adapt the material for any purpose, provided that appropriate credit is given to the original dataset creators and any changes are indicated.
+Under this license, users may share and adapt the dataset material for any purpose, provided that appropriate credit is given to the original dataset creators and any changes are indicated. The principal processing package currently does not specify a separate software license; its README describes the scope of the code release.
 
 ## Dataset Access
 
-The complete dataset is available at:
-
-https://huggingface.co/datasets/InterestingITS/U-EASWS/tree/main
+The released data and principal code are available in the [U-EASWS repository](https://huggingface.co/datasets/InterestingITS/U-EASWS/tree/main).
 
 The repository currently contains:
 
@@ -448,4 +549,5 @@ The repository currently contains:
 - [Raw_Data](https://huggingface.co/datasets/InterestingITS/U-EASWS/tree/main/Raw_Data)
 - [Basic_Trajectory_Data](https://huggingface.co/datasets/InterestingITS/U-EASWS/tree/main/Basic_Trajectory_Data)
 - [Risk_Semantic_Data](https://huggingface.co/datasets/InterestingITS/U-EASWS/tree/main/Risk_Semantic_Data)
-- [Interaction_Semantic_Data](https://huggingface.co/datasets/InterestingITS/U-EASWS/tree/main/Interaction_Semantic_Data)
+- [Interaction_Semantic_Data](https://huggingface.co/datasets/InterestingITS/U-EASWS/tree/main/Interaction_Semantic_Data)
+- [Principal Processing Code](https://huggingface.co/datasets/InterestingITS/U-EASWS/tree/main/Principal%20Processing%20Code)