Default Avatar

Ellen

Ellen

About

Username
Ellen
Joined
Visits
24
Last Active
Roles
Member

Comments

  • Hi Eric 根據step5 remove redundant layer已成功轉換onnx 不過最後inference的部份有出錯（如附圖） 我們的train只有分3類,這部份我要修改那邊參數做修正？ 謝謝. https://www.kneron.com/forum/uploads/805/JZIZ0EGEBRWA.png
    in Keras to ONNX problem Comment by Ellen May 2021
  • Hi Ethon, 因為.h5檔案太大無法上傳，提供大一點的截圖，謝謝 https://www.kneron.com/forum/uploads/239/BM9XBZFMEYKV.jpg https://www.kneron.com/forum/uploads/213/HAF5KP2H6FZ6.jpg https://www.kneron.com/forum/uploads/090/2BLPB1DMTJSC.jpg https://www.kner…
    in Keras to ONNX problem Comment by Ellen May 2021
  • Hi Quote, 關於上述回復了解train_weights_final.h5只有weight, 因此需再結合model architecture 但在yolo.py的程式碼沒有看到能將兩者合在一起的部分 我是從train.py的 ModelCheckpoint(log_dir + 'ep{epoch:03d}-loss{loss:.3f}-val_loss{val_loss:.3f}.h5',        …
    in Keras to ONNX problem Comment by Ellen May 2021
  • Hi Jiyuan Liu, 謝謝回覆  yolo example step by step 可以成功執行 您提供的tiny yolo sample（從網路上下載的weights）也可以生成 yolov3-tiny.opt.onnx 但我們根據 https://github.com/Adamdad/keras-YOLOv3-mobilenet train.py 生成的 tiny yolo model 是 tra…
    in Keras to ONNX problem Comment by Ellen April 2021
  • Hi Bike, 我們的model是根據 https://github.com/Adamdad/keras-YOLOv3-mobilenet train.py 生成的 tiny yolo model 能否套用在你們提供的 https://www.kneron.com/forum/discussion/53/example-keras-kl520-how-to-convert-and-compile-tiny-yolo-v3-from-github-pro…
    in Keras to ONNX problem Comment by Ellen March 2021
  • Hi, 參考 https://www.kneron.com/forum/discussion/53/example-keras-kl520-how-to-convert-and-compile-tiny-yolo-v3-from-github-project#latest Step 4: save weight and model architecture into one h5 file (2021/03/29 Update) this project provide us a scri…
    in Keras to ONNX problem Comment by Ellen March 2021

© Kneron, Inc.