I will train yolov8 on very small objects, tiny object detection


Vetted Pro
Vetted by Fiverr Pro
Debangsha S was selected by the Fiverr Pro team for their expertise.
About this gig
Training YOLOv8 on Very Small Objects: TINY OBJECT DETECTION!!!
Training a YOLOv8 model on a custom dataset is very easy but if your object of interest is so small that it covers less than 0.05% (for example a mouse pointer on an HD screen, satellite images, etc.) the traditional implementation of YOLO gives terrible results.
Why choose me?
- Academic qualification: I have 2 masters from the University of British Columbia, in Data Science and Computer science with a thesis focus on Computer Vision.
- Industry experience: I have over 8 years of experience working in the AI industry where I built computer vision products from scratch. Please find my profile on LinkedIn for more info.
- Timely delivery: I am a busy professional and I am sure you are too so I would try my best to deliver on time.
- Quality of the deliverable: Customer satisfaction is the ultimate priority for me so I will meet in one-on-one sessions with you to solve any queries and complaints that you might have about the final product.
Tools and Libraries:
OpenCV, PyTorch, TF, YOLO, Streamlit
NOTE: The first consultation is free. Research fee: $200 (extra). I cannot start the project without doing the research.
Get to know Debangsha S
Vetted Pro
2x CTO of AI startups, Idea to MVP in 4 to 8 weeks, AI and ML Expert
Debangsha S is part of the Fiverr Pro catalog and has been hand-picked by a dedicated Fiverr Pro team for their skills and expertise.
Vetted for
AI Consulting
AI Development
Chatbot Development
Data Analytics
Data Engineering
Data Formatting
Data Labeling & Annotation
Data Processing
Data Science & ML
Data Scraping
Data Visualization
Product Management
Project Management
Software Management
- FromCanada
- Member sinceJan 2024
- Avg. response time1 hour
- Last delivery1 year
Languages
English, Hindi, Bengali
My Portfolio
Other AI Development Services I Offer
FAQ
Does the dataset has to be of small obejcts?
Absolutely NO!!! I will annotate and train the model on any dataset you want. But if you have a small object dataset, I will do my magic.
Does the gig pricing include the initial research?
That depends. Some projects are very straightforward and don't require extra research efforts. But some projects are more difficult and they require more research. So that costs an extra $200.
Is it helpful if I am a domain expert?
Yes, it helps massively if you are a domain expert and want me to just take care of the YOLOv8 training and deployment part. In that case, we can go straight into model training and implementation after our first meeting.
Do you only do object detetction?
NO!! I will add any functionality you want to implement. Such as object segmentation (instance or class), object tracking, counting, screen thresholding etc. If it's possible with the algorithm, I will do it for you.
