The Akida Model Zoo expands our foundation models with a hand-picked collection of models accelerated by the Akida 2.0 IP. Designed for developers, researchers, and AI enthusiasts, these ready-to-use models make it easier than ever to explore, build, and innovate with the Akida solution.
Both float and quantized models are available, with quantized versions converted and evaluated on the Akida solution. For each model, the number of nodes required to run on a minimal Akida IP configuration is provided, enabling straightforward assessment of performance and deployment needs.
In addition, some models can be evaluated directly through Akida Cloud ☁️, offering a convenient way to explore and experiment without local hardware.
| Domain | Use case | Architecture | Resolution | Dataset | #Params | Quantization | Accuracy | F1 Score | MSE | Minimal #Nodes |
|---|---|---|---|---|---|---|---|---|---|---|
| Vision | Classification | MobileNetV1_1.0 | 128 | CIFAR-10 | 2.25M | 8 | 91.92% | 5 ☁️ | ||
| Vision | Classification | MobileNetV1_1.0 | 224 | Oxford_Flower | 3.3M | 8 | 91.08% | 7 | ||
| Vision | Classification | MobileNetV1 0.5 | 224 | SIIM-ISIC | 3.14M | 8 | 98.16% | 86.14% | 4 ☁️ | |
| Vision | Classification | MobileNetV1 0.5 | 224 | ODIR-5K | 3.15M | 8 | 92.00% | 97.83% | 4 ☁️ | |
| Vision | Classification | MobileNetV1 0.5 | 224 | ECG | 3.14M | 8 | 83.27% | 89.18% | 4 ☁️ | |
| Vision | Classification | MobileNetV1 0.5 | 224 | Retina OCT | 3.15M | 8 | 93.30% | 98.66% | 4 ☁️ | |
| Vision | Classification | MobileNetV2 1.0 | 224 | ImageNet | 3.5M | 8 | 70.35% | 7 | ||
| Vision | Classification | MobileNetV2 0.75 | 160 | ImageNet | 2.6M | 8 | 62.85% | 4 ☁️ | ||
| Vision | Classification | MobileNetV2 0.35 | 96 | ImageNet | 1.2M | 8 | 43.47% | 2 ☁️ | ||
| Vision | Classification | MobileNetV4 1.0 | 224 | ImageNet | 3.77M | 8 | 71.86% | 8 | ||
| Vision | Classification | MobileNetV2_1.0 | 128 | CIFAR-10 | 2.25M | 8 | 93.96% | 5 ☁️ | ||
| Vision | Classification | MobileNetV2_1.0 | 224 | Oxford_Flower | 2.4M | 8 | 91.97% | 7 | ||
| Vision | Classification | MobileNetV4_1.0 | 128 | CIFAR-10 | 2.5M | 8 | 94.72% | 7 | ||
| Vision | Classification | MobileNetV4_1.0 | 224 | Oxford_Flower | 2.6M | 8 | 85.41% | 8 | ||
| Vision | Classification | spatiotemporal | 224 | FallVision | 1.34M | 8 | 98.36% | 16 | ||
| Vision | Classification | MLP | 784 | MNIST | 203.5K | 8 | 98.05% | 1 ☁️ | ||
| Vision | Classification | LogisticRegression | 784 | MNIST | 7.9K | 8 | 84.52% | 1 ☁️ | ||
| ECG | Classification | 1DCNN | 360 | MIT-BIH | 74K | 8 | 97.3% | 1 ☁️ | ||
| ECG | Anomaly Detection | 1DCNN | 144 | ECG5000 | 290K | 8 | 94.0% | 1 ☁️ | ||
| Tabular | Classification | LogisticR. | 30 | Breast_Cancer | 169 | 8 | 93.9% | 1 ☁️ | ||
| Synthetic | Regression | MLP | 1 | 1D_Curve | 6.2K | 8 | 0.136 | 1 ☁️ | ||
| Vision | Detection | AkidaNet18/CenterNet | 224 | Soda_bottle | 2.43M | 8 | 91.53% | 6 ☁️ | ||
| Vision | Detection | AkidaNet18/CenterNet | 224 | Trail_camera | 2.43M | 8 | 84.74% | 6 ☁️ | ||
| Vision | Detection | AkidaNet18/CenterNet | 224 | Road_signs | 2.43M | 8 | 65.46% | 6 ☁️ | ||
| Vision | Detection | AkidaNet18/CenterNet | 224 | Furniture | 2.43M | 8 | 79.21% | 6 ☁️ | ||
| Vision | Detection | AkidaNet18/CenterNet | 384 | Aerial_Cows | 2.43M | 8 | 30.91% | 16 | ||
| Vision | Detection | AkidaNet18/CenterNet | 224 | Bees | 2.43M | 8 | 59.99% | 6 ☁️ | ||
| Vision | Detection | AkidaNet18/CenterNet | 224 | Cable_Damage | 2.43M | 8 | 77.32% | 6 ☁️ | ||
| Vision | Detection | AkidaNet18/CenterNet | 224 | Corrosion | 2.43M | 8 | 39.06% | 6 ☁️ | ||
| Vision | Detection | AkidaNet18/CenterNet | 224 | Digits | 2.43M | 8 | 92.32% | 6 ☁️ | ||
| Vision | Detection | AkidaNet18/CenterNet | 224 | Face_Detection | 2.43M | 8 | 75.98% | 6 ☁️ | ||
| Vision | Detection | AkidaNet18/CenterNet | 224 | Hand_Gestures | 2.43M | 8 | 53.52% | 6 ☁️ | ||
| Vision | Detection | AkidaNet18/CenterNet | 224 | License_Plate | 2.43M | 8 | 96.22% | 6 ☁️ | ||
| Vision | Detection | AkidaNet18/CenterNet | 224 | Liver_Disease | 2.43M | 8 | 40.57% | 6 ☁️ | ||
| Vision | Detection | AkidaNet18/CenterNet | 224 | Phages | 2.43M | 8 | 67.18% | 6 ☁️ | ||
| Vision | Detection | AkidaNet18/CenterNet | 224 | Sign_Language | 2.43M | 8 | 85.88% | 6 ☁️ | ||
| Vision | Detection | AkidaNet18/CenterNet | 224 | Stomata_Cells | 2.43M | 8 | 52.14% | 6 ☁️ | ||
| Vision | Detection | AkidaNet18/CenterNet | 224 | Underwater_Objects | 2.43M | 8 | 44.89% | 6 ☁️ | ||
| Vision | Detection | AkidaNet18/CenterNet | 384 | Ships_Detection | 2.43M | 8 | 39.60% | 12 | ||
| Vision | Detection | AkidaNet18/CenterNet | 224 | Bone_Fracture | 2.43M | 8 | 60.70% | 6 ☁️ | ||
| Vision | Detection | AkidaNet18/CenterNet | 224 | Facial_Expression | 2.43M | 8 | 75.40% | 6 ☁️ | ||
| Vision | Detection | AkidaNet18/CenterNet | 224 | Pothole_Detection | 2.43M | 8 | 57.20% | 6 ☁️ | ||
| Vision | Detection | AkidaNet18/CenterNet | 224 | Search_And_Rescue | 2.43M | 8 | 77.00% | 6 ☁️ | ||
| Vision | Detection | AkidaNet18/CenterNet | 224 | Traffic_Detection | 2.43M | 8 | 71.80% | 6 ☁️ | ||
| Vision | Detection | AkidaNet18/CenterNet | 384 | Weed_Crop | 2.43M | 8 | 47.70% | 16 |
To avoid downloading the models during cloning due to their large size:
GIT_LFS_SKIP_SMUDGE=1 git clone git@github.com:Brainchip-Inc/models.gitTo download a specific model:
git lfs pull --include="[path to model]" --exclude=""To download all models:
git lfs pull --include="*" --exclude=""Alternatively, you can download models directly from GitHub. Navigate to the model's page and click the "Download" button on the top right corner.
For a graphical representation of each model's architecture, we recommend using Netron.