READ ME File For 'Dataset for thesis "Perception of Natural Scenes: Depth Estimation and the Influence of Colour"' Dataset DOI: 10.5258/SOTON/PG/D067 ReadMe Author: Michaela Trescakova, University of Southampton, ORCID ID 0000-0002-2194-1243 This dataset supports the thesis entitled "Perception of Natural Scenes: Depth Estimation and the Influence of Colour" AWARDED BY: University of Southampton DATE OF AWARD: 2026 License: Creative Commons Attrribution (CC-BY) Date of data collection: 2020-2026 Information about geographic location of data collection: UK Date that the file was created: 11.6.2026 -------------------- DATA & FILE OVERVIEW -------------------- This dataset contains data and materials used for thesis "Perception of Natural Scenes: Depth Estimation and the Influence of Colour" , spanning chapters 2, 3 and 4: 'Chapter_2.zip' 'Ch2_Exp1.csv' - Participant perception data collected for Experiment 1 'Ch2_Exp2.csv' - Participant perception data collected for Experiment 2 'PIS.docx' - Participant Information Sheet for Experiment 2 'Consent Form.docx' - Participant Consent Form for Experiment 2 'Chapter_3.zip' 'Ch3_raw_points.csv' - Participant segmentation data collected for generation of segmentation dataset 'Ch3_Segm.zip' - Segmentation dataset used for the experiment: inside zip file are 76 '.pkl' files, each is an image mask composite denoting segmented areas of an image. 'Ch3_Exp.csv' - Participant perception data collected for Experiment 'Chapter_4.zip' 'Ch4_Exp1.csv' - Participant perception data collected for Experiment 1 'Ch4_Exp2.csv' - Participant perception data collected for Experiment 2 'Ch4_Exp3.csv' - Participant perception data collected for Experiment 3 'Ch4_Exp4.csv' - Participant perception data collected for Experiment 4 -------------------------- METHODOLOGICAL INFORMATION -------------------------- Details of data collection/generation are described in detail in the thesis, here is a brief overview: 'Ch2_Exp1_P.csv': raw data collected via online platform JATOS - participants from Prolific or SONA conducted an online experiment of a depth estimation task, where their responses were collected. 'Ch2_Exp2.csv': collected in stereoscopic vision laboratory in University of Southampton, Building 44 - participants conducted an online experiment of a depth estimation task, where their responses were collected. 'Ch3_raw_points.csv': raw data collected via online platform JATOS - participants drew points on images and labelled objects in the images, coordinates and labels were collected. 'Ch3_Segm.zip': 76 segmentation mask files generated from the 'Ch3_raw_points.csv', by processing them through Segment Anything Model-2 model (https://ai.meta.com/research/sam2/) and edited by the researcher. .pkl files can be inspected using Python, pickle library. Each segmentation file is associated with a defined SYNS image scene (https://syns.soton.ac.uk/), defined in metadata. 'Ch3_Exp.csv': raw data collected via online platform JATOS - participants conducted an online experiment of a depth estimation task, where their responses were collected. 'Ch4_Exp1.csv': raw data collected via online platforms JATOS and Qualtrics - participants conducted an online experiment where they judged whether images contain certain objects(animal, building, ...). 'Ch4_Exp2.csv': raw data collected via online platforms JATOS and Qualtrics - participants conducted an online experiment where they judged whether images contain certain objects(animal, building, ...). 'Ch4_Exp3.csv': raw data collected via online platforms JATOS and Qualtrics - participants conducted an online experiment where they judged whether images contain certain objects(animal, building, ...). 'Ch4_Exp4.csv': raw data collected via online platforms JATOS and Qualtrics - participants conducted an online experiment where they judged whether images contain certain objects(animal, building, ...). -------------------------- DATA-SPECIFIC INFORMATION -------------------------- Here is content of each data file - column and row meanings: 'Ch2_Exp1.csv': each row represents 1 trial . | Column | Meaning | ----------------------------- | -------------------------------------------------------------------------------------------------------- | `scene` | SYNS Scene ID. | `view` | SYNS View ID within the scene. | `elevation` | Elevation condition - 1 = same, 2 = incongruent, 3 = congruent. | `p1_row` | Row pixel coordinate of point 1. | `p1_col` | Column pixel coordinate of point 1. | `p1_dist` | Depth/distance value at point 1. | `p1_meandist` | Mean depth around point 1, local patch mean. | `p1_stddist` | Standard deviation of depth around point 1. | `p2_row` | Row pixel coordinate of point 2. | `p2_col` | Column pixel coordinate of point 2. | `p2_dist` | Depth/distance value at point 2. | `p2_meandist` | Mean depth around point 2. | `p2_stddist` | Standard deviation of depth around point 2. | `meandist` | Experimental depth-distance condition: 10, 20, 40. | `distcontrast` | Depth contrast condition, coded 1=small span around the mean depth, 3= large span around mean depth. | `comb_std` | Combined depth variability for the two points, apparently `p1_stddist + p2_stddist`. | `fixdisp` | Central fixation disparity/displacement in depth . | `colour` | Colour condition =1 = natural, 2 = greyscale, 3 = unnatural. | `scene_cat` | Scene category code- SYNS category. | `nat_vs_built` | Natural vs built environment code. | `p1_side` | Side on which point 1 was displayed, coded 2=left, 1 = right. | `correct_side` | Correct response side, coded 2=left, 1 = right. | `depth_difference` | Absolute depth difference between point 1 and point 2. | `mean_depth` | Mean of `p1_dist` and `p2_dist`. | `correct_perc` | Correct percentage response, `p1_dist / p2_dist × 100`. | `X` | Participant index. If < 100, it was SONA participant, if not then Prolific. | `perc_response` | Participant’s percentage estimate/response. | `side_response` | Participant’s side choice, coded 2=left, 1 = right. | `prestime` | Presentation time condition: 2= 33 ms, 6=100 ms, 16 = 267 ms. | `rt` | Reaction time, in milliseconds. | `col1` | Answer to 1st Ishihara colour plate test, 15 is correct. | `col2` | Answer to 1st Ishihara colour plate test, 8 is correct. | `col3` | Answer to 1st Ishihara colour plate test, 42 is correct. | `prolific` | Prolific/SONA ID. | `OS` | Participant operating system. | `browser` | Participant web browser. | `age` | Participant age. | `sex` | Participant sex/gender entry. | `focus` | Attention/focus score or self-report. | `RTtotal` | Total experiment duration. | `Is_correct_side` | Whether `side_response` matched `correct_side`; 1 = correct, 0 = incorrect. | `perc_deviation_from_correct` | Absolute deviation between `perc_response` and `correct_perc`; missing when side response was incorrect. 'Ch2_Exp2.csv': each row represents 1 trial . | Column | Meaning | ----------------------------- | ---------------------------------------------------------------------------------------------------------------------------- | `scene` | SYNS scene ID. | `view` | SYNS view ID within the scene. | `elevations` | Elevation condition: `1` = same, `2` = incongruent, `3` = congruent. Same as `elevation` in Exp1 | `p1_row` | Row pixel coordinate of point 1. | `p1_col` | Column pixel coordinate of point 1. | `p1_dist` | Depth/distance value at point 1. | `p1_meandist` | Mean local depth around point 1. | `p1_stddist` | Standard deviation of local depth around point 1. | `p2_row` | Row pixel coordinate of point 2. | `p2_col` | Column pixel coordinate of point 2. | `p2_dist` | Depth/distance value at point 2. | `p2_meandist` | Mean local depth around point 2. | `p2_stddist` | Standard deviation of local depth around point 2. | `meandepths` | Experimental mean-depth condition: `10`, `20`, or `40`. Same as `meandist` in Exp1. | `depthcontrasts` | Depth contrast condition: `1` = small span around mean depth, `3` = large span. Same as `distcontrast` in Exp1. | `comb_std` | Combined local depth variability for the two points, `p1_stddist + p2_stddist`. | `fixdisp` | Central fixation disparity/displacement in depth. | `stereo` | Stereo presentation condition - 1 = monocular, 2 = binocular presentation. | `colours` | Colour condition: `1` = natural, `2` = greyscale/BW, `3` = unnatural. Same as `colour` in Exp1 | `path` | File path to the stimulus folder. Encodes elevation, colour, mean-depth, and contrast-bin conditions. | `img` | Stimulus image filename, e.g. `Scene54_View7.png`. | `prestimeframes` | Presentation duration in frames: `2`, `6`, or `16`, corresponding to 33 ms, 100 ms, and 267 ms. | `rt` | Reaction time / response duration, seconds . | `emp_prestime` | Empirical/actual presentation time in seconds | `whichfurther` | Participant’s side choice for which point was further, `2` = left, `1` = right | `percentresponse` | Participant’s percentage estimate/response. Same as `perc_response` in Exp1. | `participant` | Participant ID / initials. | `p1_side` | Side on which point 1 was displayed: `2` = left, `1` = right, as in Exp1. | `correct_side` | Correct side response: `2` = left, `1` = right. | `Is_correct_side` | Whether `whichfurther` matched `correct_side`; `1` = correct, `0` = incorrect. | `correct_perc` | Correct percentage response, calculated as `p1_dist / p2_dist × 100`. | `perc_deviation_from_correct` | Absolute deviation between `percentresponse` and `correct_perc`; missing when side response was incorrect. | `depth_difference` | Absolute depth difference between point 1 and point 2. | `mean_depth` | Mean of `p1_dist` and `p2_dist`. | `cond` | Condition/block code - needed for the 3×3 Latin-square rotation across triads of participants (details in the thesis) 'Ch3_raw_points.csv': each row represents 1 point clicked on an image. | Column | Meaning | --------------- | ----------------------------------------------------------------------------------------------------------- | `ParticipantID` | Participant / annotator ID. | `Color` | Hex colour assigned to the labelled object/region during annotation. | `X` | Raw x-coordinate of the point on the displayed annotation canvas. | `Y` | Raw y-coordinate of the point on the displayed annotation canvas. | `Image` | Source image filename, e.g. `31_Im16.jpg`, where 31 means SYNS scene, and 16 means SYNS view for that scene | `Label` | Object label assigned to the annotated point, e.g. `glazing bar`, `sky`, `gravel`, `tree`. | `Prolific` | Prolific participant ID. | `CanvasWidth` | Width of the displayed annotation canvas in pixels. | `CanvasHeight` | Height of the displayed annotation canvas in pixels. | `x_rescaled` | X-coordinate rescaled from the displayed canvas to the original image size. | `y_rescaled` | Y-coordinate rescaled from the displayed canvas to the original image size. 'Ch3_Segm.zip' 76 .pkl files can be inspected using Python, pickle library. Each pickle file contains composite segmentation masks (4000 × 6000 pixels each) stored as integer label masks, and associated per-object metadata. Each row of metadata corresponds to one segmented area/object in the SYNS scene. The columns are as follows: | Metadata column | Meaning | ---------------------------- | ------------------------------------------------------------------------------------------------------------ | `participant` | Participant ID. | `image` | Source image filename. for example: `38_Im2.jpg`, which corresponts to SYNS scerne 38, view 2. | `mask_index` | Segment ID used inside the composite mask. Pixels with this integer belong to that object/region. | `color` | Hex colour used to display that segment in annotation/visualisation. | `label` | Object/material/category label, e.g. Brick, Fence, Bush, Pipe. | `mask_quality` | Original mask quality rating, 1 = best, 5 = worst. | `mask_idx` | Choice of initial mask form SAM2 output - it proivides 3 versions to pick from, so 1-3. | `new_mask_quality` | Mask qualtiy rating after manual edits 1 = best, 5 = worst. | `label_rating` | Quality/confidence rating for the assigned label. | `splitting_needed` | Flag indicating whether the segment should be split into smaller objects/regions. | `repeat` | Flag indicating a repeated/duplicate object or repeated annotation. | `size` | Segment area in pixels. Same as `mask_pixel_count`. | `width` | Width of the segment bounding box, in pixels. | `height` | Height of the segment bounding box, in pixels. | `depth_min` | Minimum depth value within the segment. | `depth_max` | Maximum depth value within the segment. | `depth_mean` | Mean depth value within the segment. | `depth_median` | Median depth value within the segment. | `depth_mode` | Most frequent or modal depth value within the segment. | `depth_std` | Standard deviation of depth values within the segment. | `depth_q1` | First quartile depth value. | `depth_q2` | Second quartile depth value, essentially the median. | `depth_q3` | Third quartile depth value. | `depth_valid_pct` | Percentage of mask pixels with valid depth values. | `depth_valid_count` | Number of pixels in the mask with valid depth values. | `mask_pixel_count` | Total number of pixels in the segment mask. | `depth_min_center70` | Minimum depth in the central 70% region of the segment. | `depth_max_center70` | Maximum depth in the central 70% region. | `depth_mean_center70` | Mean depth in the central 70% region. | `depth_median_center70` | Median depth in the central 70% region. | `depth_mode_center70` | Modal depth in the central 70% region. | `depth_std_center70` | Standard deviation of depth in the central 70% region. | `depth_q1_center70` | First quartile depth in the central 70% region. | `depth_q2_center70` | Second quartile / median depth in the central 70% region. | `depth_q3_center70` | Third quartile depth in the central 70% region. | `depth_valid_pct_center70` | Percentage of valid depth pixels in the central 70% region. | `depth_valid_count_center70` | Number of valid depth pixels in the central 70% region. | `mask_pixel_count_center70` | Number of pixels included in the central 70% region. 'Ch3_Exp.csv':each row represents 1 trial . | Column | Meaning | ----------------------------- | -------------------------------------------------------------------------------------------------------------------------------------------------------- | `scene` | SYNS scene ID. | `view` | SYNS view ID within the scene. | `categorytype` | Object manipulation, values `1=manmade-manmade`, `2=nature-nature`, `3=mixed`. The filename also contains this as `cat1`, `cat2`, or `cat3`. | `p1_row` | Row pixel coordinate of point 1. | `p1_col` | Column pixel coordinate of point 1. | `p1_dist` | Depth/distance value at point 1. | `p1_meandist` | Mean local depth around point 1. | `p1_stddist` | Standard deviation of local depth around point 1. | `p2_row` | Row pixel coordinate of point 2. | `p2_col` | Column pixel coordinate of point 2. | `p2_dist` | Depth/distance value at point 2. | `p2_meandist` | Mean local depth around point 2. | `p2_stddist` | Standard deviation of local depth around point 2. | `meandist` | Experimental mean-depth condition, here `15m` or `30m`. | `distcontrast` | Depth contrast condition: `1` = small span around mean depth, `2` = large span | `filename` | Stimulus image filename; encodes scene, view, category type, mean depth, and depth contrast. | `X` | Participant index. | `perc_response` | Participant’s percentage estimate/response. | `side_response` | Participant’s side choice, `2` = left and `1` = right, as in previous files. | `prestime` | Presentation time condition in frames: `2`, `6`, or `16`, corresponding to 33 ms, 100 ms, and 267 ms. | `col1` | Answer to first Ishihara colour plate test; `15` is correct. | `col2` | Answer to second Ishihara colour plate test; `8` is correct. | `col3` | Answer to third Ishihara colour plate test; `42` is correct. | `rt` | Reaction time in milliseconds. | `ID` | Prolific ID. | `OS` | Participant operating system. | `browser` | Participant web browser. | `age` | Participant age. | `sex` | Participant sex/gender entry. | `col` | Colour- 1 = natural, 2 = greyscale | `focus` | Attention/focus score or self-report. | `RTtotal` | Total experiment duration, in milliseconds. | `prestime2` | Recorded/actual presentation duration in frames . | `correct_side` | Correct response side, `2` = left and `1` = right. | `correct_perc` | Correct percentage response, calculated as `p1_dist / p2_dist × 100`. | `Is_correct_side` | Whether `side_response` matched `correct_side`; `1` = correct, `0` = incorrect. | `perc_deviation_from_correct` | Absolute deviation between `perc_response` and `correct_perc`; missing when side response was incorrect. 'Ch4_Exp1.csv':each row represents 1 trial . | Column | Meaning | ----------------------- | ------------------------------------------------------------------------------------------------------------------- | `Type_key` | Abbreviated target stimulus category, e.g. `Animal`, `Build`, `Dry`, `For`, `Mount`. | `Trial_num` | Trial/item number within category, values `1`–`4` -there were 4 trials per each stimulus category | `Q_num` | Numeric response selected by participant, values `1`–`4`. Same as `Q`, with a few missing values. | `selected_raw` | Raw selected category label, lower-case / sometimes abbreviated, e.g. `animal`, `build`, `dry`, `for`. | `StartDate` | Qualtrics session start date/time, stored as numeric date-time. | `EndDate` | Qualtrics session end date/time. | `Duration (in seconds)` | Total survey duration in seconds. | `Finished` | Whether participant completed the survey; `1` = finished, `0` = not finished. | `Prolific ID` | Prolific participant ID. | `gender` | Participant gender response, coded numerically, 1 = man, 1 = woman, 3 = other. | `age` | Participant age. | `coldef` | colour-vision self-report response 1 = deficient, 2 = not deficient, tested, 3 = not deficient, never tested. | `test1_rt_First Click` | Time to first click for catch/test question 1. | `test1_rt_Last Click` | Time to last click for test question 1. | `test1_rt_Page Submit` | Page-submit time for test question 1. | `test1_rt_Click Count` | Number of clicks for test question 1. | `Q-tes1` | Response to test question 1. | `C-test1_1` | Confidence rating for test question 1, 0–100. | `test2_rt_First Click` | Time to first click for test question 2. | `test2_rt_Last Click` | Time to last click for test question 2. | `test2_rt_Page Submit` | Page-submit time for test question 2. | `test2_rt_Click Count` | Number of clicks for test question 2. | `Q-test2` | Response to test question 2. | `C-test2_1` | Confidence rating for test question 2, 0–100. | `test3_rt_First Click` | Time to first click for test question 3. | `test3_rt_Last Click` | Time to last click for test question 3. | `test3_rt_Page Submit` | Page-submit time for test question 3. | `test3_rt_Click Count` | Number of clicks for test question 3. | `Q-test3` | Response to test question 3. | `C-test3_1` | Confidence rating for test question 3, 0–100. | `id` | Participant ID - prolific or sona. | `Group` | Experimental image condition: `original`= natural colour sharp, `original_BW`= greyscale sharp, | | `typical` = natural colour blur, `rotated = rotated blur`, `greyscale`=greyscale blur, `pro`= protan blur | | `tri` = tritan blur. | `Type` | Full target stimulus category: `Animal`, `Beach`, `Building`, `Dryland`, `Field`, `Forest`, `Fruit`, or `Mountain`. | `Trial` | Trial/item number within category, values `1`–`4`. Same as `Trial_num`. | `First` | Time to first click on the current stimulus trial. | `Last` | Time to last click on the current stimulus trial. | `Page` | Page-submit time for the current stimulus trial. | `Click` | Number of clicks on the current stimulus trial. | `Q` | Participant’s numeric response on the current stimulus trial, values `1`–`4`. | `C` | Participant confidence rating for the current stimulus trial, 0–100. | `correct` | Correct numeric response for the current stimulus trial. | `is_correct` | Accuracy for the current trial; `1` = correct, `0` = incorrect. | `selected_stimulus` | Cleaned label of the participant’s selected category. | `Stimulus` | Stimulus identity, combining category and item number, e.g. `Animal1`, `Beach3`, `Mountain4`. 'Ch4_Exp2.csv': each row represents 1 trial . | Column | Meaning | -----------------------| ------------------------------------------------------------------------------------------------------------------- | `X` | Participant index. | `cond` | Image condition, e.g. `grey_blur`, `nat_normal`, `nat_blur`, `rotated`(meaning rotated blur), `rotated_normal`, `protan`(meaning protan blur), `protan_normal`, | | `tritan`(meaning tritan blur), `tritan_normal`, `grey_normal`. 'normal' meaning sharp. | `prolific` | Prolific participant ID. | `age` | Participant age. | `sex` | Participant sex/gender entry, typed by participant . | `col_choice` | Self-report response about colour-vision status/testing. | `col8` | Answer to Ishihara colour plate with correct answer `8`. | `col15` | Answer to Ishihara colour plate with correct answer `15`. | `col42` | Answer to Ishihara colour plate with correct answer `42`. | `practice_response` | Stimulus selected during the practice trial. | `practice_response_rt` | Response time for selecting the practice stimulus. | `practice_slider` | Slider/confidence response for the practice trial. | `practice_slider_rt` | Response time for the practice slider selection . | `response` | Stimulus image selected on the current trial. | `response_rt` | Response time for selecting the stimulus on the current trial. | `slider` | Slider/confidence response for the current trial, ranging from 0 to 100. | `slider_rt` | Response time for the current-trial slider. | `submit_rt` | Page-submit time for the current trial. | `prompt` | Target category prompt shown to the participant, e.g. `FIELD`, `MOUNTAIN`, `FRUIT`, `PERSON`, `FOREST`, `WATER`, `ANIMAL`, `SHIP`, `FLOWER`, `CAR`, last 3(SHIP,FLOWER,CAR) are catch trials. | `correct` | Accuracy of the selected response; `1` = correct, `0` = incorrect. | `detected` | Category detected/assigned from the selected stimulus image, e.g. `field`, `fruit`, `human`, `forest`, `mountain`, `water`, `animal`; `NA` for catch trials. 'Ch4_Exp3.csv':each row represents 1 trial. | Column | Meaning | ---------------------- | ----------------------------------------------------------------------------------------------------------------------------------- | `X` | Participant index. | `cond` | Image condition. | `prolific` | Prolific participant ID. | `age` | Participant age. | `sex` | Participant sex/gender entry. | `col_choice` | Self-report response about colour-vision status/testing. | `col8` | Answer to Ishihara colour plate with correct answer `8`. | `col15` | Answer to Ishihara colour plate with correct answer `15`. | `col42` | Answer to Ishihara colour plate with correct answer `42`. | `practice_response` | Stimulus selected during the practice trial. | `practice_response_rt` | Response time for selecting the practice stimulus. | `practice_slider` | Slider/confidence response for the practice trial. | `practice_slider_rt` | Response time for the practice slider. | `response` | Stimulus image selected on the current trial. | `response_rt` | Response time for selecting the stimulus on the current trial. | `slider` | Slider/confidence response for the current trial, ranging from 0 to 100. | `slider_rt` | Response time for the current-trial slider. | `submit_rt` | Page-submit time for the current trial, in milliseconds. | `prompt` | Target category prompt shown to the participant: `ANIMAL`, `FIELD`, `FOREST`, `FRUIT`, `MOUNTAIN`, `PERSON`, or `WATER`. | `correct` | Accuracy of the selected response; `1` = correct, `0` = incorrect. | `detected` | Category detected/assigned from the selected stimulus image: `animal`, `field`, `forest`, `fruit`, `human`, `mountain`, or `water`. | `submit_rt_sec` | Page-submit time converted from milliseconds to seconds. | `outlier_rt` | Reaction-time outlier flag; `1` = outlier, `0` = retained trial. | `submit_rt_log` | Log-transformed page-submit time. 'Ch4_Exp4.csv':each row represents 1 trial. | Column | Meaning | --------------- | ------------------------------------------------------------------------------------------------------------------- | `X` | Participant index. | `cond_target` | Image condition of the target stimulus. | `cond_response` | Images conditions of the response window. | `prolific` | participant ID. | `age` | Participant age. | `sex` | Participant sex/gender entry. | `col_choice` | Self-report response about colour-vision status/testing. | `col8` | Answer to Ishihara colour plate with correct answer `8`. | `col15` | Answer to Ishihara colour plate with correct answer `15`. | `col42` | Answer to Ishihara colour plate with correct answer `42`. | `response` | File path of the stimulus image selected by the participant. | `response_rt` | Response time for selecting the stimulus image. | `slider` | Slider/confidence response for the current trial, ranging from 0 to 100. | `slider_rt` | Response time for the slider/confidence response. | `submit_rt` | Page-submit time for the current trial, in milliseconds. | `target` | File path of the target stimulus image shown/promoted on the trial. | `prompt` | Target category for the trial: `animal`, `fruit`, `human`, or `mountain`. | `detected` | Category detected/assigned from the participant’s selected response image. | `cond_col` | Colour/image-transformation of the trial : `nat`, `grey`, `rotated`, `protan`, or `tritan`. | `cond_blur` | Blur pairing between target and response conditions: `blur_normal`, `normal_blur`, `blur_blur`, or `normal_normal`. | `correct` | Accuracy of the selected response; `1` = correct, `0` = incorrect. | `cond_col_blur` | Combined condition label made from `cond_col` and `cond_blur`.