How to Use
1. Upload Drone Files
Upload the following required files: DJI drone video, CSV file, SRT file All files must be uploaded to proceed to the next page.

File Upload Policy
Maximum File Sizes:
Videos: Up to 5GB
Images: Up to 1GB
Other files: Up to 500MB
Minimum Requirements:
The video must be at least 5 seconds long.
The resolution must be FHD (1920x1080) or higher.
Metadata must be present in the file; analysis cannot proceed without it.
✅ DJI Drone Data File Conversion Guide
To use data captured by a DJI drone, you need to convert the extracted text (.txt) file into a CSV file. Click the [TXT to CSV Conversion Guide] button to access the conversion site and step-by-step instructions.

Additional Tip
Ensure that the DJI Flight Record data matches the recorded video. To avoid data mismatches, carefully verify the flight record when capturing footage.
2. Enter Shooting Information
Fill in the required shooting details to proceed with the analysis.

Required Information
The information fields may vary. Fields marked with (*) are mandatory and must be filled in to proceed to the next page.
3. Select AI Analysis Section
Use the bottom slider to adjust the analysis range. Green-marked sections indicate frames with high distance estimation accuracy.

Section Length Limit
The selected section must be at least 5 seconds long.
The maximum length allowed is 1 minute.
✅ Guide to Additional AI Analysis Features
Enhance AI analysis with advanced AI algorithms. Enabling additional features may increase processing time.
Glare Removal: Reduces excessive light reflections in the video to improve analysis accuracy.
False Detection Reduction (Non-Sea Areas): Helps distinguish non-sea areas, minimizing false detections of objects outside the ocean.

✅ Submit AI Analysis Request
Click the [Start AI Analysis] button to submit your request. You will receive a confirmation email with your request ID once the AI analysis request is successfully submitted.
AI Analysis Processing Time
The analysis may take some time to complete.
Once the AI analysis is finished, you will receive an email with the results.
Didn't receive the email? Check your spam folder.
4. Select Frames with Boat Markers
DIVA estimates the distance between dolphins and boats based on the boats' known length to determine potential violations. For each section, select frames where a boat with a known length appears by clicking the [ ✔ ] button.

If No Object Is Detected
If the object is not detected, select another frame where a boat with a known length appears. This helps ensure accurate distance estimation.

Managing Multiple Segments
If the AI assigns different unique IDs to the same object, multiple segments may be created.
Use the [Eraser] button to modify or remove previously selected frames.
5. Enter Boat Length Information
The boat's length is used to determine potential violations. Adjust the endpoints of the displayed line on the screen to match the actual length of the boat.

Tips for Accurate BoatLength Input
Drag the line to adjust its position.
Move the endpoints to match the boat’s actual full length.
Double-check the measured length before confirming.
Entering more boat length data improves distance estimation accuracy.
✅ Boat Image Information
In the [Boat Image Information] tab, you can view images and length details of boats that frequently approach dolphins. If you don’t know the boat’s length, check this tab for reference.

Using the Service Without Boat Length Information
You can still use the service even if you don’t know the boat’s length, but distance measurement accuracy may be lower.
✅ AI Violation Results Generation
Click the [Generate Results] button to start processing the violation analysis. You will receive emails notifying you of the processing status and completion of the results.
AI Violation Results Processing Time
The analysis may take some time to complete.
Once the violation results are generated, you will receive an email with the analysis report.
Didn’t receive the email? Check your spam folder.
7. Review Violation Results
Check the distance between the boat and dolphins to determine any violation based on the analysis results.

Violation Probability Indicator
If the violation probability is high, the distance color will be displayed in red.
Detailed visualized information can be found in [Legend] Violation Results Visualization.
✅ Violation Timeline
Use the timestamps to quickly navigate to the boat violation sections.

✅ Data Storage Format
AI analysis results are saved in multiple formats, allowing you to review detailed information, including violation status.
a. Raw Data Storage
The AI analysis results are stored frame by frame.
Filename format: {fileName}_rawdata.csv
Tab_tracking results
date/time
Date and time when the analysis results were saved
frame
Frame number of analyzed video
object_id
Unique Identifier assgined to each object analyaed by AI
class
Type of Object Analyzed by AI (e.g., boat, dolphin)
bbox_xn, bbox_y
Coordinates of the BBox Detected by AI
(x1, y1), (x2, y2), (x3, y3), (x4, y4)
speed_knots
Speed of the detected boat in the analyzed frame (kt)
distance_m
Distance between the detected dolphin and boat (m)
Tab_illegal results
frame
Frame number of analyzed video
cases
Illegal Code
_ If illegal, indicate the code number (e.g., [2], [2,3])
_ If not illegal, leave it blank (e.g., [ ])
Illegal Case Codes:
_ Case1: Distance between dolphin and vessel within 50m
_ Case2: Vessel movement within 50-300m
_ Case3: Speed exceeding 5kt within 300-750m
_ Case4: Speed exceeding 10kt within 750-1500m
_ Case5: Three or more vessels within 300m
illegal
Violation Status
_ If illegal, TRUE
_ If not illegal, FALSE
b. Save Current Frame
Saves the frame at the paused moment in the video
Filename format: {fileName}_frame_{frame_ID}_results.jpg
c. Save General View Video
Saves the perspective-applied video file
Filename format: {fileName}_viz_results.mp4
d. Save Top-Down(Bird's Eye View) Video
Converst the perspective-applied image into an orthogonally projected (no-perspective) video and saves it.
Filename format: {fileName}_bev_viz_results.mp4
Last updated