Basic Amateur Soccer Match Analysis Algorthym Developing
Budget: $3,000 – $5,000 USD
I'm seeking a skilled computer vision expert to develop an object detection and recognition system specifically for detecting and recognizing persons. This system will be used in both indoor and outdoor environments.
Key requirements:
- High accuracy in diverse settings
- Real-time processing capability
- Robustness to varying lighting and weather conditions
Ideal skills and experience:
- Proficiency in Python and OpenCV
- Experience with deep learning frameworks (TensorFlow, PyTorch)
- Background in developing and deploying computer vision models
My name is Furkan, and I am a practicing lawyer based in Turkey. As part of my work, I provide legal consultancy to several sports facilities.
For quite some time, we have been trying — and investing in — the development of a computer vision-based software that can analyze individual and team performance in 7v7 or 8v8 hobby football matches. Unfortunately, we have not yet been able to achieve meaningful or consistent results.
I recently came across your work, and it truly excited me. That’s why I wanted to reach out personally.
I’m sharing with you several match recordings from our football fields, taken from different camera angles. I would love to know if you think we can work together to at least develop a system that provides basic statistics, supported by a reliable player tracking algorithm — if it is technically feasible.
There are some physical limitations in our fields that I’d like to mention:
There is no clearance space between the field and the perimeter.
The entire field is surrounded by wire mesh fencing.
At about 6 meters in height, there is also a net installed across the top.
Fine-tune a custom object detection model if necessary
Develop an accurate and fast object tracking algorithm
Implement team classification
Apply homographic transformation
Initially focus on extracting only player and team-related statistics (excluding the ball), such as:
Total distance covered
Sprints
Estimated calories burned
Heatmaps
Corner kicks
Penalties
Fouls, etc.
If we decide to proceed with a panoramic camera instead of using two separate cameras, we can install the 16MP 4K Reolink Duo 3 Panoramic Camera.
Matches are about 1 hour long
Despite these challenges, we are fully committed to the project and ready to make any necessary improvements to enhance video capture quality.
If, after reviewing the footage I’ve sent, you believe that such a solution is technically achievable, I would greatly appreciate it if you could also share a rough price estimate for the development work.
Looking forward to hearing your thoughts.
Kind regards,
Furkan
Key requirements:
- High accuracy in diverse settings
- Real-time processing capability
- Robustness to varying lighting and weather conditions
Ideal skills and experience:
- Proficiency in Python and OpenCV
- Experience with deep learning frameworks (TensorFlow, PyTorch)
- Background in developing and deploying computer vision models
My name is Furkan, and I am a practicing lawyer based in Turkey. As part of my work, I provide legal consultancy to several sports facilities.
For quite some time, we have been trying — and investing in — the development of a computer vision-based software that can analyze individual and team performance in 7v7 or 8v8 hobby football matches. Unfortunately, we have not yet been able to achieve meaningful or consistent results.
I recently came across your work, and it truly excited me. That’s why I wanted to reach out personally.
I’m sharing with you several match recordings from our football fields, taken from different camera angles. I would love to know if you think we can work together to at least develop a system that provides basic statistics, supported by a reliable player tracking algorithm — if it is technically feasible.
There are some physical limitations in our fields that I’d like to mention:
There is no clearance space between the field and the perimeter.
The entire field is surrounded by wire mesh fencing.
At about 6 meters in height, there is also a net installed across the top.
Fine-tune a custom object detection model if necessary
Develop an accurate and fast object tracking algorithm
Implement team classification
Apply homographic transformation
Initially focus on extracting only player and team-related statistics (excluding the ball), such as:
Total distance covered
Sprints
Estimated calories burned
Heatmaps
Corner kicks
Penalties
Fouls, etc.
If we decide to proceed with a panoramic camera instead of using two separate cameras, we can install the 16MP 4K Reolink Duo 3 Panoramic Camera.
Matches are about 1 hour long
Despite these challenges, we are fully committed to the project and ready to make any necessary improvements to enhance video capture quality.
If, after reviewing the footage I’ve sent, you believe that such a solution is technically achievable, I would greatly appreciate it if you could also share a rough price estimate for the development work.
Looking forward to hearing your thoughts.
Kind regards,
Furkan