This README.txt file was generated on 2022-11-10 by MALIKA NISAL RATNAYAKE.

GENERAL INFORMATION

1. Title of Dataset: Spatial Monitoring and Insect Behavioural Analysis Dataset

2. Author Information
	A. MALIKA NISAL RATNAYAKE (Email: malika.ratnayake@monash.edu)

	B. DON CHATHURIKA AMARATHUNGA (Email: don.amarathunga@monash.edu)

	C. AZADUZ ZAMAN (Email: asaduzzaman@monash.edu)

	D. ADRIAN G DYER (Email: adrian.dyer@rmit.edu.au, adrian.dyer@monash.edu)

	E. ALAN DORIN (Email: alan.dorin@monash.edu)
	
3. Date of data collection: 2021-03-09 to 2021-03-17 

4. Geographic data collection location: Sunny Ridge Strawberry Farm, Boneo, Australia.


SHARING/ACCESS INFORMATION

1. Associated Publication authors: M.N. Ratnayake, D.C. Amarathunga, A. Zaman, A.G.Dyer and A. Dorin

2. Associated Publication: "Spatial Monitoring and Insect Behavioural Analysis Using Computer Vision for Precision Pollination.", Accepted to be published in the International Journal of Computer Vision (Acceptance date: 7th November 2022)

3.Recommended citation for this dataset: Ratnayake, Malika Nisal; Amarathunga, Don Chathurika; Zaman, Azaduz; Dyer, Adrian G; Dorin, Alan (2022): Spatial Monitoring and Insect Behavioural Analysis Dataset. Monash University. Dataset. https://doi.org/10.26180/21533760


DATA & FILE OVERVIEW

1. File Arrangement:

Spatial Monitoring and Insect Behavioural Analysis Dataset
 |- Test_Video_Dataset.zip*(1) => Test Video Dataset used for Experimental Evaluation of the Automated Multi-Species Tracking Algorithm (Please refer to the associated publication for more information)
 |- YOLOv4_Training_and_Test_Dataset*(2,3) => Annotated datasets for training and testing YOLOv4 model
 | |- classes.names => Names of classes
 | |- training => Training dataset with annotations. 
 | |- testing => Test dataset with annotations.
 |- Insect_Trajectory_Dataset.zip*(4,5) => Insect and flower trajectory dataset with associated pollination analysis and browsing tools
 | |- Graphs => Folder containing Pollination Analysis plots from the associated publication (Fig. 5-8). Generated using Pollination_Analyser.ipynb.
 | | |- data_summary.png
 | | |- species_contributions.png
 | | |- temporal_data.png
 | | |- track_lengths.png
 | |- Pollination_Analysis_Summary => Summary data related to pollination analysis generated using the Pollination_Analyser.ipynb 
 | | |- all_flowers.csv
 | | |- all_tracks.csv
 | | |- data_point_summary.csv
 | | |- flower_coverage.csv
 | | |- flower_summary.csv
 | | |- flower_visits.csv
 | | |- position_list.txt
 | | |- time_summary.csv
 | | |- track_list.txt
 | | |- track_summary.csv
 | |- core => Methods for pollination analysis
 | | |- Track_Browser_Methods.ipynb
 | | |- Track_Processor_Methods.ipynb
 | |- Extracted_Tracks*(6) => Tracks extracted and processed by the Polytrack Software**
 | | |- Flower_Positions_CSV*(7) => Folder containing final positions of recorded flowers
 | | |- Flower_Tracks_CSV*(8) => Folder containing flower trajectories
 | | |- Insect_Images_PNG*(9) => Folder containing images of recorded insects
 | | |- Insect_Tracks_CSV*(10) => Folder containing extracted insect trajectories
 | |- Interactive_Track_Visualiser.ipynb => Jupyter Notebook to browse and visualise individual insect tracks and tracks recorded at each data collection point
 | |- Pollination_Analyser.ipynb => Jupyter Notebook to analyse tracks and flower positions to extract pollination-related data and plot graphs (Fig. 5-8 in publication).
 |- README.txt

*(1) File name convention for videos: Cam_N_T_YYYYMMDD_HHMMSS.h264.avi, where N = Camera Number, T = Heading direction of the top corner of the video (North/South), YYYYMMDD = Recorded Year/Month/Date, and HHSSMM = Video start time HH:MM:SS. (e.g. cam_4_N_video_20210309_132604.h264.avi: Camera = 4, Heading direction of top corner of the video = North, Record date = 2021/03/09, Record time = 13:26:04)

*(2) Ref https://github.com/theAIGuysCode/YOLOv4-Cloud-Tutorial for YOLOv4 Object detector training procedure.

*(3) Number of instances per class in training and testing datasets.
Honeybees/Vespidae=> Training: 2231/371 instances | Testing: 258/44 instances
Starwberry Flower=> Training: 14050 instances | Testing: 2909 instances
Syrphidae=> Training: 204 instances | Testing: 20 instances
Lepidoptera=> Training: 93 instances | Testing: 15 instances

*(4) Relevant codes used to extract tracks and analyse data can be found at https://github.com/malikaratnayake/Polytrack2.0. 

*(5) The associated video dataset is available upon request. Email malika.ratnayake@monash.edu or alan.dorin@monash.edu to request video data.

*(6) Distribution of number of tracks in the dataset:
Honeybees: 1805
Syrphidae: 85 
Lepidoptera: 100
Vespids: 345
Starwberry Flowers: 379

*(7) File name convention for flower positions: flowers_PDD.csv, where P = Data collection point (1-9, see Fig.2b of the associated publication), and DD = Recorded date. (e.g. flowers_108.csv: Data collection Point = 1, Recorded date = 08th March)

*(8) File name convention for flower tracks: flowes_tracks_PDD.csv, where P = Data collection point (1-9, see Fig.2b of the associated publication), and DD is the recorded date. (e.g. flowes_tracks_108.csv: Data collection Point = 1, Recorded date = 08th March)

*(9) File name convention for insect images: PDDHHMMSS0I.png, where P = Data collection point (1-9, see Fig.2b of the associated publication), DD = Recorded date, HHMMSS = Recorded time of the day HH:MM:SS, and I = Insect type (0 = Honeybee, 1 = Syrphidae, 2= Lepidoptera, 3 = Vespidae) (e.g. 10913474603.png: Data collection point = 1 , Recorded date = 09th March, Recorded time of the day = 13:47:46, and Insect type = 3 - Vespidae)  

*(10) File name convention for insect tracks: PDDHHMMSS0I.csv, where P = Data collection point (1-9, see Fig.2b of the associated publication), DD = Recorded date, HHMMSS = Recorded time of the day HH:MM:SS, and I = Insect type (0 = Honeybee, 1 = Syrphidae, 2= Lepidoptera, 3 = Vespidae) (e.g. 10913474603.csv: Data collection point = 1 , Recorded date = 09th March, Recorded time of the day = 13:47:46, and Insect type = 3 - Vespidae) 

Note: Additional related data collected that was not included in the current data package: https://github.com/malikaratnayake/Polytrack2.0

METHODOLOGICAL INFORMATION

1. Description of methods used for collection/generation of data: Information on data collection can be found in the associated publication.

2. Methods for processing the data and pollination analysis: CSV and Video files of insect tracks were obtained using Polytrack2.0 (https://github.com/malikaratnayake/Polytrack2.0)

DATA-SPECIFIC INFORMATION FOR: Insect_Tracks_CSV folder

Organisation of the CSV files:
nframe_InsectID: Recorded frame number of the insect
absframe_InsectID: Frame number in the day
x0_InsectID: X coordinate (pixels)
y0_InsectID: Y coordinate (pixels)
area_InsectID: Area covered by the insect (pixels)
yadj_InsectID: Adjusted Y coordinate (pixels) - considers the top edge of the video is towards the North.
flower_InsectID: Visited flower number
visit_num_InsectID: Visit number to the specific flower from the current track (1=first visit, 2=second visit...)
sx_InsectID: Smoothed X coordinate (pixels)
sy_InsectID: Smoothed Y (Adjusted) coordinate (pixels)
speeds_InsectID: Speed of the insect

DATA-SPECIFIC INFORMATION FOR: Flower_Positions_CSV and Flower_Tracks_CSV folders

Organisation of the CSV files:
flower_num: Assigned number of the flower
x0: X coordinate of the center (pixels)
y0: Y coordinate of the center (pixels)
y_adj: Adjusted Y coordinate of the center (pixels) - considers the top edge of the video is towards the North.
radius: Flower radius
occlusion: Number of frames insect is occluded
species: Detection class determined by YOLOv4
confidence: YOLOv4 detection confidence
Total_time: Total time (no.of frames) insects spent on the flower (only for Flower_Positions_CSV)
Total_visits: Total insect visits to the flower (only for Flower_Positions_CSV)