Maria Chen
Maria Chen
About
- Username
- Maria Chen
- Joined
- Visits
- 652
- Last Active
- Roles
- Member, ModeratorsKL520, ModeratorsKL720, ModeratorsAIModel
Comments
-
Hi Haresh, The error says, "Input name does not match," with the onnx input ({input_nodes}) and given np ({input_np.keys()}), so you might want to make sure that the order of the names are matching the order of inputs in pre_results: https…
-
Hi, Unfortunately, the natural language processing sample isn't available to the public. If you were interested, please contact our sales rep. Thank you!
-
Hi Haresh, There might be a problem with the onnx file. Could you provide your onnx file so we could test it out and look into it? Thanks! You could also check if your onnx model has operators supported by KL720: Hardware Supported Operators - Docum…
-
您好, 您可以先確認您的model的operators都是有支援的: Hardware Supported Operators - Document Center (kneron.com) 之後,我們建議您用onnx2onnx.py來做optimization和轉換成Kneron形式的onnx檔。
-
Hi Haresh, For now, KL630 is only available through sales channels and it's not publicly available yet. Please feel free to contact our sales department or the department you made a purchase from for technical support.
-
您好, 1. Kneron Academy只支援KL520裝置: https://www.kneron.com/forum/uploads/788/HJXD3LQLABR4.png Kneron Academy是幾年前的app,比較沒有在維護了,建議您使用Kneron PLUS。 2. 只要segmentation模型內的operators都是KL720有support的,理論上KL720裝置就能支…
-
Hi Jacky, 太好了,恭喜您轉換與推論成功!
-
Hi Jacky, 不好意思,回覆變晚了! 關於kneron_flow (0216)的model,我們model team的同事在onnx檔發現post process的部分留在onnx裡面了。Post process高機率會包含KL720裡面的NPU不支援的operator,所以會有異常。我們建議您將onnx裡面的post process的部分(下圖,紅…
-
Hi Jacky, 新年快樂, 關於yolov5 to kneron.nef的流程,(4)是Kneron格式的.onnx to .bie,(5)是.bie to .nef沒錯 是的,(3)的onnx to onnx, (4), 和(5)都需要在Toolchain Docker上去執行。您可以參照這裡的步驟,將python code的部分放在一個python script裡面(像是to…
-
Hi Jacky, Toolchain的workspace/libs/compiler裡面有kneron_nef_utils,可以查看.nef模型的資訊: NEF Utils Guide - Document Center (kneron.com) 不過沒有辦法看到詳細的架構。
-
Hi Jacky, 我們沒有yolo v5最新版本的轉換教學,不過舊的版本有這個範例: Object detection yolov5 - Document Center (kneron.com) ,可能會需要與這個新的toolchain manual一起做參考: 1. Toolchain Manual Overview - Document Center (kneron.com) 無論是任何版本…
-
您好, KL630的開發流程也很類似KL520/720的流程,所以理論上客戶不會有太大的問題 PLUS推薦目前的最新版本,裡面有KL630的範例: v2.2.0 docker也推薦目前的最新版本: v0.23.1 KL630應用範例: Create Single Model Example for KL630 - Document Center (kneron.com)
-
Hi, The kp_release_model_nef_descriptor function only releases the struct of the model information, so it doesn't unload the model on the device. If the device already has a model, we can skip kp_load_model and run inference. If you'd like to load a…
-
您好, 您可以參照文檔上在Raspberry Pi設定KL520和KL720裝置的說明: Install Dependency - Document Center (kneron.com) 之後,可以確認裝置的firmware是KDP2的: Upgrade AI Device To KDP2 - Document Center (kneron.com) 設定好之後,可以使用Kneron PLUS,用裡面…
-
Hi Tri, It might be hard to debug with only the information above. Could you provide us with your NEF files and the program file you're running, so we could look into them? Thanks!
-
Hi Tri, The NEF files can't be combined because the versions of the NEF files are different: one is v0, and the other one is v1. You could check the versions of the NEF files with: ./kneron_nef_utils --info path/to/file.nef To resolve this, you coul…
-
Hi Nhan, When running multiple models, you would need to reboot the KL720 dongle. The size limit for models in KL720 is 75MB when uploading model via USB and 70MB when loading model from flash: https://www.kneron.com/forum/uploads/272/BXBKQBAXK0GU.p…
-
Hi Roy, 我們這邊沒有對外公開RI2019.2的版本,若有專案需求,請您跟直接向專案窗口詢問,謝謝。
-
Hi Haresh, Here are the steps to set up and use KL720 dongle on your Linux, with documentation links to guide you through: Download Kneron PLUS and Kneron DFUT (under the Kneron PLUS section): Developers | Kneron – Full Stack Edge AI Install Depende…
-
Hi Nhan, From what I'm seeing on Netron (Netron is a website where you could visualize models), it seems like both onnx models in 7f134165 and 6020f8a6 have operators that are not supported on KL720, such as Squeeze and Unsqueeze. https://www.kneron…
-
Hi Nhan, Could you provide the onnx model as well? Thank you! One thing you could try running is the IP Evaluator: 3. Floating-Point Model Preparation - Document Center (kneron.com) eval_result = km.evaluate() Printing out eval_result would show wha…
-
您好, 是的,要把識別的物體的數據(inference的結果)發給另一個MCU,會需要自行加上IIC通訊,還有模型需要用我們NCPU支援的模型,理論上就可以發過去了。
-
您好, 除非有用自己的模型的需求,一般來說NCPU不需要做改動。您要跑HICO Camera Inference Demo的話,可以直接使用在PLUS裡面已有的fw_ncpu.bin,不需要額外去build NCPU。 如果要編輯或build NCPU,會需要跟我們的業務窗口申請開放權限,也需要有DSP (Xtensa Xplore…
-
您好, KL720 SDK裡面有包含SCPU與NCPU,但是NCPU的project位置並不是直接放在solution_kdp2_hico_mipi裡面。
-
您好, 是的,KL720的firmware有SCPU和NCPU。KL720 SDK的solution_kdp2_hico_mipi裡面SCPU和NCPU都有包含。Kneron PLUS的res/firmware裡面已有的fw_scpu.bin和fw_ncpu.bin都是在KL720 SDK裡面build好的。 Kneron PLUS是一個我們可以用Kneron的產品跑inference的介面,…
-
您好, KL720 SDK的solution_kdp2_hico_mipi與Kneron PLUS Enterprise的HICO Camera Inference demo是相對應的。 https://www.kneron.com/forum/uploads/002/EDAHYD3P0KET.png 要跑HICO Camera Inference,會需要build在KL720 SDK裡面的solution_kdp2_hico_mipi,生成f…
-
您好, 是的,會需要Hico MIPI camera。Firmware的SCPU需要自行去build,而NCPU照理上可以直接使用Kneron PLUS裡面的NCPU。 我們有Kneron PLUS Enterprise (含有Hico MIPI的demo code) 開放給特定的客戶,如果有需要開放權限可以向我們的業務窗口詢問與申請。 Enterpri…
-
您好, 是的,Kneron PLUS是通過PC的USB做輸入的。 如果後續有任何問題,請再跟我們說,謝謝!
-
您好, 要跑solution_kdp2_hico_mipi,會需要準備相對應的KL720板子(KL720_DMB_BGA386_SCHEMATIC_VOP2)與Hico MIPI camera。 如果沒有Hico MIPI camera,可以先跑Kneron PLUS的基本範例。Kneron PLUS有提供範例模型可以直接使用。 您可以在Kneron的網站下載Kneron PLUS…
-
Hi Wing, (1 , 33 ,76 , 76), (1 , 33 ,38 , 38), and (1 , 33 , 19 , 19) seem like output shapes of the onnx model. To visualize your onnx model, you could use this website: https://netron.app/ Could you describe the following in detail so we could und…