Face Recognition, Image Classification, Image Enhancement…
Is your smartphone capable of running the latest Deep Neural Networks to perform these AI-based tasks? Does it have a dedicated AI Chip? Is it fast enough? Run AI Benchmark to comprehensively evaluate it's AI Performance!
AI Benchmark measures the speed, accuracy and memory requirements for several key AI and Computer Vision algorithms. Among the tested solutions are Image Classification and Face Recognition methods, Neural Networks used for Image Super-Resolution and Photo Enhancement, AI models playing Atari Games and performing Bokeh Simulation, as well as algorithms used in autonomous driving systems. Visualization of the algorithms’ output allows to assess their results graphically and to get to know the current state-of-the-art in various AI fields.
In total, AI Benchmark consists of 21 tests and 11 sections provided below:
Section 1. Classification, MobileNet-V2
Section 2. Classification, Inception-V3
Section 3. Face Recognition, Inception-ResNet-V1
Section 4. Playing Atari Games, LSTM
Section 5. Deblurring, SRCNN
Section 6. Super-Resolution, VGG19
Section 7. Super-Resolution, SRGAN
Section 8. Bokeh Simulation, U-Net
Section 9. Semantic Segmentation, ICNet
Section 10. Image Enhancement, DPED ResNet
Section 11. Memory limits, SRCNN
Note: Hardware acceleration is supported on Android 9.0 and above on all mobile SoCs with AI accelerators, including Qualcomm Snapdragon, HiSilicon Kirin, Samsung Exynos and MediaTek Helio.
1. New Tasks: OCR, Text Completion and Parallel Model Execution.
2. NNAPI-1.2 compatible Neural Networks added.
3. The total number of tests increased to 46.
4. Native hardware acceleration on Snapdragon and Exynos SoCs using the Hexagon NN / Eden delegates.
5. Extended accuracy measurements.
6. DSP / NPU throttling tests (PRO mode).
7. Running custom TFLite models (PRO mode).
8. Optimizations for low-RAM devices.
9. GPU-based AI acceleration is available on devices with OpenGL ES 3.0+ support.