Focoos Models 🧠#
With the Focoos SDK, you can take advantage of a collection of foundational models that are optimized for a range of computer vision tasks. These pre-trained models, covering detection and semantic segmentation across various domains, provide an excellent starting point for your specific use case. Whether you need to fine-tune for custom requirements or adapt them to your application, these models offer a solid foundation to accelerate your development process.
Semantic Segmentation 🖼️#
| Model Name | Architecture | Domain (Classes) | Dataset | Metric | FPS Nvidia-T4 |
|---|---|---|---|---|---|
| fai-m2f-l-ade | Mask2Former (Resnet-101) | Common Scene (150) | ADE20K | mIoU: 48.27 mAcc: 62.15 |
73 |
| fai-m2f-m-ade | Mask2Former (STDC-2) | Common Scene (150) | ADE20K | mIoU: 45.32 mACC: 57.75 |
127 |
| fai-m2f-s-ade | Mask2Former (STDC-1) | Common Scene (150) | ADE20K | mIoU: 41.23 mAcc: 52.21 |
189 |
mIoU = Intersection over Union averaged by class
mAcc = Pixel Accuracy averaged by class
FPS = Frames per second computed using TensorRT with resolution 640x640
Object Detection 🕵️♂️#
| Model Name | Architecture | Domain (Classes) | Dataset | Metric | FPS Nvidia-T4 |
|---|---|---|---|---|---|
| fai-rtdetr-l-coco | RT-DETR (Resnet-50) | Common Objects (80) | COCO | bbox/AP: 53.06 bbox/AP50: 70.91 |
87 |
| fai-rtdetr-m-coco | RT-DETR (STDC-2) | Common Objects (80) | COCO | bbox/AP: 44.69 bbox/AP50: 61.63 |
181 |
| fai-rtdetr-s-coco | RT-DETR (STDC-1) | Common Objects (80) | COCO | bbox/AP: 42.58 bbox/AP50: 59.22 |
220 |
| fai-rtdetr-n-coco | RT-DETR (STDC-1) | Common Objects (80) | COCO | bbox/AP: 40.59 bbox/AP50: 56.69 |
269 |
| fai-rtdetr-m-obj365 | RT-DETR (Resnet50) | Common Objects (365) | Objects365 | bbox/AP: 34.60 bbox/AP50: 45.81 |
87 |
AP = Average Precision averaged by class
AP50 = Average Precision at IoU threshold 0.50 averaged by class
FPS = Frames per second computed using TensorRT with resolution 640x640
Instance Segmentation 🎭#
| Model Name | Architecture | Domain (Classes) | Dataset | Metric | FPS Nvidia-T4 |
|---|---|---|---|---|---|
| fai-m2f-l-coco-ins | Mask2Former (Resnet-50) | Common Objects (80) | COCO | segm/AP: 42.39 segm/AP50: 66.12 |
54 |
AP = Average Precision averaged by class
AP50 = Average Precision at IoU threshold 0.50 averaged by class
FPS = Frames per second computed using TensorRT with resolution 640x640