Livestock counting software project

Develop a private web-based platform that can automatically count cattle in drone-captured photos or videos, with an estimated error margin of 4–6%, and allow the user to manually correct any mistakes before exporting the final results in PDF format.


---

Key Features

File Upload

Direct upload of images (jpg/png)

direct upload of videos (mp4/mov, full hd 1080p)


> this system does not work in real time while the drone is flying. The video or images must be recorded first, and then uploaded to the platform.


Automatic Animal Detection (GPU Acceleration)

AI model detects each animal

Shows total count

Generates an image/frame with each animal numbered

Fast processing thanks to a cloud server with GPU

Manual Editing

User can add, remove, or adjust detections

Total count updates instantly after corrections

Helps fix cases where the system misses an animal or counts one twice


Exportable Results

PDF report including:

Annotated image/frame

Final corrected count

Date and basic analysis info


CSV export optional


Private Access

Login with username and password

All data and results remain private for the client

---

User Flow

1. Drone records photos or video


2. User uploads the file to the platform


3. System analyzes and displays automatic counting


4. User corrects detections if necessary


5. Final PDF report is generated and downloaded

---

Suggested Technology Stack

Frontend: React / Next.js

Backend: FastAPI (Python) or Node.js

AI Model: yolo-style animal detection optimized for gpu

infrastructure: cloud server with gpu (t4/l4 recommended)

storage: s3 or similar bucket storage

database: postgresql

expected accuracy

target error margin: 3–5%

accuracy improves over time with real samples provided by the client


deliverables

✔ fully working private web platform
✔ image and video upload
✔ automatic cattle counting with visual annotations
✔ manual correction tools
✔ pdf and csv export
✔ secure private login
✔ basic documentation + usage guide.


Optional Add-ons (quoted separately)

WhatsApp or email notifications when processing is finished

Field/paddock labeling

Future module for weight estimation (video only)

Custom branding (logo, colors, domain)