125 points by objectdetection_enthusiast 5 months ago flag hide 12 comments
johnsmith 5 months ago next
Great job! I've been following this project and it's impressive to see it come together.
creator 5 months ago next
Thanks for the kind words, excited to share more updates soon!
creator 5 months ago next
We used GPU-acceleration and implemented optimizations such as pruning and quantization to improve frames per second (FPS).
otheruser 5 months ago prev next
Can you share more details about the machine learning models used?
creator 5 months ago next
Sure! We mainly used YOLOv3 for real-time object detection.
helpfulperson 5 months ago prev next
I found this tutorial on ML models to be helpful, hopefully it helps you too!
otheruser 5 months ago next
Thank you for sharing, I'll check it out!
neuralnetwork_follower 5 months ago prev next
Did you experiment with other architectures like EfficientDet or Faster R-CNN?
creator 5 months ago next
Yes, we briefly explored those models but YOLOv3 proved to be the best fit for real-time requirements.
fan 5 months ago prev next
Seeing the progress, any plans on open-sourcing this project?
creator 5 months ago next
Absolutely! We're dedicated to collaborating with the HN community and other developers. Stay tuned!