search

Seeed-Studio / sscma-example-esp32

Example of SenseCraft Model Assistant Model deployment related to ESP32
22 stars 12 forks source link

The source train code of model provided in "model zoo" #19

Open shupinghu opened 7 months ago

shupinghu commented 7 months ago

By infer the "yolov5_tiny_face.tflite"(the input size is 192x192) model provided in "model zoo", the time cost is 360ms, but by infer the model provided in "https://github.com/Seeed-Studio/sscma-model-zoo/blob/main/docs/en/person_Detection_Swift-YOLO_192.md"(the input size is 192x192 as well, and it's structure is yolov5_tiny as well), the time cost is 600ms.

The reason is that he model structure of "yolov5_tiny_face.tflite" is different with normal yolov5 structure, would you please provide the structure code of "yolov5_tiny_face.tflite" in "https://github.com/Seeed-Studio/ModelAssistant"?

LynnL4 commented 7 months ago

Hi, Yes, yolov5_tiny_face uses a different network structure, I'll let you know when I upload the specific network configuration structure

shupinghu commented 6 months ago

Hi, Yes, yolov5_tiny_face uses a different network structure, I'll let you know when I upload the specific network configuration structure

Thank you very much. It seems that the specific network configuration has not uploaded until today, could you please tell me the definite release time?

LynnL4 commented 5 months ago

Hi, sorry, we've lost this config file and may not be able to provide it. can you try 96x96 swift-yolo?