i am only half way through the video and this is by far the best one i've seen yet. zero issues following, very clearly explained and the structure is perfect. done! Thanks!!: MBP M2 Pro on 640x640 video running at 29ms per frame on battery power, 20ms on full power. Rough calculation with ms per 100k pixels shows it's about 4x faster than M1. Also 170ms on CPU is quite good
Yoooo your blog was so short and accurate! Didn't know that "mps" is the equivalent to cuda to using the mac gpu! My processtimes reduced from 3000ms to 100ms! Lessgo
Can the teacher guide a lesson to identify objects with yolo and control the webcam to follow that object? using jetson nano. Thank you very much sir, have a nice evening
On my M2 MacBook Air I got this result (in line with the older GPU?) 0: 384x640 1 dog, 28.6ms Speed: 1.4ms preprocess, 28.6ms inference, 11.0ms postprocess per image at shape (1, 3, 384, 640)
Hi, thanks your video. When I use my macbook pro (M1pro) test my MPS is appear true, but run my code I can't saw the speed is incerase. It seems like doesn't work by gpu. How can I to solve ? tks
Hello Pysource, can you please provide an idea or tutorial to predict a person weight using computer vision in real time ? Can you please help me on that
Hi On my laptop Rtx 3060, with input video 1280x720 and inference image size 800x480, i get 6ms with Tensor Rt, segemtation medium model. Wich is near 10 times faster than M1... And on the M1, post-process is 13.4ms...
Are you using miniconda M1 version? if so which check the python using mps like below: import platform print(platform.platform()) you should get output something like this -- "macOS-13.2.1-arm64-arm-64bit"
Hi, i am searching for a tool to identify archaeological finds from photos. We need size, color, and hopefully with a big database, type, and date when it was used. Do you know any?