Elon Musk
@elonmusk
En réponse à @samiralyateem
All frames stitched to single frame (tricky, as all cameras overlap in different ways) & then creating video segments for labeling & training. Everything from the labeling tools themselves to training to inference had to be rewritten.
15:37 · 12 septembre 2020 · 0 vues
51
65
847