AI Benchmark

Ignatov Andrey

Mayelana nalolu hlelo lokusebenza

Ukubonwa kobuso, ukuhlukaniswa kwesithombe, ukuphendulwa kwemibuzo...

Ingabe i-smartphone yakho iyakwazi ukusebenzisa ama-Deep Neural Networks akamuva ukwenza le misebenzi neminye eminingi esekelwe ku-AI? Ingabe inayo i-AI Chip ezinikele? Ingabe ishesha ngokwanele? Qalisa i-AI Benchmark ukuze uhlole ngobungcweti Ukusebenza kwayo kwe-AI!

Izinga lefoni lamanje: http://ai-benchmark.com/ranking

I-AI Benchmark ikala isivinini, ukunemba, ukusetshenziswa kwamandla kanye nezidingo zenkumbulo zokhiye abambalwa be-AI kanye ne-Computer Vision algorithms. Phakathi kwezixazululo ezihloliwe kukhona Ukuhlelwa Kwezithombe kanye nezindlela Zokuqashelwa Kobuso, Amanethiwekhi E-Neural asetshenziselwa I-Image/Video Super-Resolution kanye Nokuthuthukisa Isithombe, amamodeli e-AI abikezela umbhalo nokuphendula imibuzo, kanye nezixazululo ze-AI ezisetshenziswa ezinhlelweni zokushayela ezizimele nama-smartphones ngokoqobo- Isikhathi Ukujula Ukulinganisa kanye Semantic Image Segmentation. Ukubonakala kwemiphumela yama-algorithms kuvumela ukuhlola imiphumela yawo ngemidwebo kanye nokwazi isimo samanje sezobuciko emikhakheni ehlukahlukene ye-AI.

Sekukonke, i-AI Benchmark iqukethe izivivinyo ezingama-78 nezigaba ezingama-26 ezibalwe ngezansi:

Isigaba 1. Ukuhlelwa, i-MobileNet-V2
Isigaba 2. Ukuhlelwa, Ukuqaliswa-V3
Isigaba 3. Ukubonwa kobuso, i-MobileNet-V3
Isigaba 4. Ukuhlelwa, i-EfficientNet-B4
Izigaba 5/6. Ukwenziwa Kwemodeli Ehambisanayo, 8 x Ukuqalwa-V3
Isigaba 7. Ukulandelwa Kwezinto, YOLO-V4
Isigaba 8. Ukubonwa Kwezinhlamvu Zokubona, i-CRNN
Isigaba 9. Isegimenti Semantic, DeepLabV3+
Isigaba 10. I-Parallel Segmentation, 2 x DeepLabV3+
Isigaba 11. Ukufiphaliswa Kwezithombe, IMDN
Isigaba 12. Image Super-Resolution, ESRGAN
Isigaba 13. Image Super-Resolution, SRGAN
Isigaba 14. Ukwehliswa kwezithombe, i-U-Net
Isigaba 15. Ukulinganisa Ukujula, i-MV3-Ukujula
Isigaba 16. Ukuthuthukiswa Kwesithombe, I-DPED ResNet
Isigaba 17. Ukuthuthukiswa Kwesithombe, I-DPED Instance
Isigaba 18. I-Bokeh Effect Rendering, i-PyNET+
Isigaba 19. I-ISP yekhamera efundiwe, iPUNET
Isigaba 20. I-FullHD Video Super-Resolution, XLSR
Isigaba 21/22. I-4K Video Super-Resolution, VideoSR
Isigaba 23. Ukugcwaliswa Kombhalo, LSTM
Isigaba 24. Ukuphendulwa Kwemibuzo, MobileBERT
Isigaba 25. Ukugcwaliswa Kombhalo, ALBERT
Isigaba 26. Imikhawulo Yenkumbulo, ResNet

Ngaphandle kwalokho, umuntu angalayisha futhi ahlole amamodeli abo okufunda okujulile e-TensorFlow Lite Kumodi ye-PRO.

Incazelo enemininingwane yokuhlolwa ingatholakala lapha: http://ai-benchmark.com/tests.html

Qaphela: Ukusheshiswa kwezingxenyekazi zekhompyutha kusekelwa kuwo wonke ama-SoC eselula anama-NPU azinikele nama-accelerator e-AI, okuhlanganisa i-Qualcomm Snapdragon, i-HiSilicon Kirin, i-Samsung Exynos, i-MediaTek Helio / Dimensity kanye nama-chipset e-UNISOC Tiger. Kusukela ku-AI Benchmark v4, umuntu angakwazi futhi ukunika amandla ukusheshisa kwe-AI okusekelwe ku-GPU kumadivayisi amadala kuzilungiselelo ("Sheshisa" -> "Vumela Ukusheshisa kwe-GPU", i-OpenGL ES-3.0+ iyadingeka).

Yini entsha

1. Updated Qualcomm QNN and MediaTek Neuron delegates.
2. Enhanced stability and accuracy of the power consumption test.
3. Various bug fixes and performance improvements.