Maria Chen
Maria Chen
About
- Username
- Maria Chen
- Joined
- Visits
- 1,195
- Last Active
- Roles
- Member, ModeratorsKL520, ModeratorsKL720, ModeratorsAIModel
Comments
-
您好， 這個模型看起來有Squeeze這個operator，而這個operator我們的device不支援: https://www.kneron.com/forum/uploads/655/NAMAR0PLJLC0.png 請參照這個表格，並且確認模型的operators都是KL720會支援的: Hardware Supported Operators - Document Center
-
Hi Jyoti, Thank you for explaining your application. For further assistance, please contact us through email, looping our sales representative as well.
-
Hi Jyoti, Once you use KL630 to run RTSPs and streamer, you could stream from the PC side by using GStreamer or other software. We haven't tried using GStreamer with KL630 yet; could you explain in detail what exactly you'd like to achieve with livi…
-
您好， 感謝您提供資訊。NNM的inference速度確實會比Kneron PLUS+USB快，不過直接使用NNM之前請先確認在Kneron PLUS的KL730 Single Model Example也可以跑，在NNM做deployment會比較順利。 Yolov5的模型有很多種，不同的模型會搭配不同的postprocess，所以訓練出來的y…
-
您好， Error code 21是KP_ERROR_INVALID_MODEL_21，意思是模型不符合規範/不正確，或是model ID路徑錯誤 請問一下: 1. yolov5_211_model_730.nef是指既有的KNERON_YOLOV5S_COCO80_640_640_3 (model ID = 211)模型嗎? 2. 您的KL730 SDK的版本是哪一版的呢? 3. 您使用Kn…
-
Hi Francis, 1. We can provide you with the IMX678 driver, but the source code for it is usually for internal use, and you'll need to consult with our sales team for that. https://www.kneron.com/forum/uploads/494/LL5MLDPMF86M.zip 2. The Device Tree/D…
-
您好， 之前在Kneron Toolchain上有combine_nef的模型跑失敗，會遇到KP_FW_INFERENCE_ERROR_101的bug，kneron_nef_utils可能也有相同的bug。這個在Toolchain的最新版本(v0.31.0)有修好，可以請您更新Toolchain版本再來試一次。 也可以請您說明一下"無法進行推論&q…
-
您好， 也謝謝您回報錯誤，Toolchain現在release了最新版本(v0.31.0)，ktc.combine_nef的問題有修正。
-
Hi, Apologies for the late reply. What kind of SD card are you using? If you're using a SanDisk SD card, there's a high possibility it won't work. https://www.kneron.com/forum/uploads/525/OK0T4WGHQFEQ.png Further info on booting KNEO Pi: Boot KNEO P…
-
您好， 這個問題重複了，關於模型的問題請前往這個連結，謝謝: yolov5_211_model_730.nef 是否有提供修改過yolov5s 模型Source Code — Kneron Developer Forums
-
您好， 不好意思，我們沒有修改過yolov5s模型的source code。我們的開發中心是有yolov7的modification說明，但是沒有yolov5s的。 https://www.kneron.com/forum/uploads/406/YXQGCCE77XHC.png 可以請您說明一下nef模型無法直接使用的原因嗎? 是指nef結果不佳，還是用ne…
-
您好， 在使用Kneron PLUS測試KL720的nef模型之前，可以先使用Toolchain上的E2E Simulator來比較您的onnx模型，bie模型，以及nef模型的inference結果 (onnx模型也可以去用ONNX Runtime去跑)，檢查是從哪一個模型開始會有不正常的結果。 參考文檔: 5. Compilation - Doc…
-
您好， 謝謝您提供資訊! 我們發現Toolchain版本v0.28.2之後的ktc.combine_nef有問題，這個會在未來的Toolchain版本修正。如果要使用ktc.combine_nef，會需要請您退版到v0.28.1的Toolchain去使用ktc.combine_nef: docker pull kneron/toolchain:v0.28.1
-
您好， 從model_description.txt看不太出問題，可以請您提供以下的檔案及資訊，讓我們復現問題嗎? 謝謝! -Kneron PLUS code -Kneron PLUS版本 -分別的兩個模型，A.nef和B.nef -Combined後的A_B.nef -是否有修改firmware的code呢?
-
Hi, 很高興問題有解決! 是的，用DFUT_Console的時候跟用Kneron DFUT的指令相似，都要用sudo。 Kneron DFUT 範例: https://www.kneron.com/forum/uploads/307/9Q1ATRP05ZC6.png
-
您好， 有一點點偏差是正常的，不過如果偏差太大的話，您可以使用toolchain上的E2E simulator來比較onnx，bie，和nef模型的推論結果。若onnx和nef的結果差太多的話可以調整量化的參數。 請參照這篇文章: [FAQ] 用NEF推論不如預期 / NEF inference results are differen…
-
您好， 請問一下，您的firmware版本和Kneron PLUS版本是什麼呢? 因為firmware版本是2的話有可能會出現問題，所以會需要您更新firmware。還有，可以請您確認在Raspberry Pi5之外的device都可以認得到M.2嗎? 謝謝! 因為雖然介面是M.2，連接方式是用USB，所以僅認得到USB …
-
您好， 不好意思，我們沒有M.2相關的文檔，不過只要電腦可以讀取M.2，就可以在Kneron PLUS上使用，使用方式和USB dongle一樣喔。
-
您好， 這是因為host_stream和companion_user_ex (Kneron PLUS)的inputs不一樣，host_stream的input是從sensor來的。host_stream 中預設串流進來的影像格式為 yuv40p，須要設定為 KP_IMAGE_FORMAT_YUV420，而設定為 KP_IMAGE_FORMAT_RAW8 能運行是因為hardware只取了灰…
-
您好， 不是的，image_format的設定是用來告訴硬體輸入影像的色彩空間，並不會去轉換影像 1. 可以請您測試一下這個KL630DemoGenericImageInferenceTesting.py，切換true/false的flag來測試看看並提供一下結果與生成的RGB/gray binary file嗎? https://www.kneron.com/f…
-
您好， 謝謝您提供更多資訊。如果連KL630DemoGenericImageInference.py也會出問題的話，我們可以先解決這個問題，然後再去應用到KL630 solution_host_stream上。 KL630DemoGenericImageInference.py裡面的kp.GenericInputNodeImage，可以設定normalization，default為K…
-
您好， Int16模型的錯誤，是因為16-bit模型目前不支持hardware image pre-process，所以需要您自行做前處理 (請參考kl630_demo_generic_data_inference.c)。 關於模型預測需要先改成灰階再使用KP_IMAGE_FORMAT_RGB565的問題，可以請您提供NEF模型，normalization設定，…
-
Hi Zhihao, Yes, you're using the right conda environment for opset 11. The Resize operator isn't supported by KL520 so it should be removed and added in the postprocess function. Unfortunately, we could usually cut off operators that are not weighed…
-
Hi Zhihao, What is your onnx model's opset? If it was opset 13 or higher, we'd recommend you to use the onnx1.13 environment inside the toolchain. Also, please check the following: -Your onnx model has gone through onnx2onnx optimization -Your onnx …
-
Hi, Thank you for providing the log. For the code, did you mean the one below? https://www.kneron.com/forum/uploads/666/MJZF4ULWRPE2.png If so, the Yolov3 example on our documentation is outdated; it was written for toolchain v0.22.0, and our latest…
-
Hi Dongkyun, What is showing up on your UART log? Please make sure that your KL730 firmware is the one used for Kneron PLUS. The file location is: SDK\04_Firmware\ubifs\fw_plus_nand_4k.ubifs As for flashing firmware, please refer to this documentat…
-
Hi Sathish, Sorry for the late reply. For questions on KL730, please contact the sales representative you purchased the KL730 from, as they would provide you with the contact person for technical assistance. Thank you!
-
Hi Sathish, Could you provide your models and code so we could take a better look at this issue? For questions on KL730, you could also contact the sales representative you purchased the KL730 from. They should be able to provide you with the conta…
-
Hi Sathish, Unfortunately, we can't run models parallely in NPU, but if you were in companion mode and used 2 dongles, it could perform nearly as close as running parallely.
-
Hi, I was talking about flash_image_solution_host_mipi.bin earlier, but for the flash_image_solution_hico_mipi.bin you were trying, you will need to have access to Kneron PLUS enterprise, then you could build and run [kneron_plus]/examples_enterpris…