91、使用paddleocr V5进行算能开发板适配
基本思想：使用paddleocr V5进行算能开发板测试 技术资料
一、登录开发板进行基础信息获取
import platform
import sys
import os
import time
import threadinglive_ip = 0def get_os():os = platform.system()if os == "Windows":return "n"else:return "c"def ping_ip(ip_str):cmd = ["ping", "-{op}".format(op=get_os()),"1", ip_str]output = os.popen(" ".join(cmd)).readlines()for line in output:if str(line).upper().find("TTL") >= 0:print("ip: %s is ok ***" % ip_str)global live_iplive_ip += 1breakdef find_ip(ip_prefix):''''' 给出当前的127.0.0 ，然后扫描整个段所有地址 '''threads = []for i in range(1, 256):ip = '%s.%s' % (ip_prefix, i)threads.append(threading.Thread(target=ping_ip, args={ip, }))for i in threads:i.start()for i in threads:i.join()if __name__ == "__main__":print("start time %s" % time.ctime())cmd_args = sys.argv[1:]args = "".join(cmd_args)ip_pre = "192.168.10"find_ip(ip_pre)print("end time %s" % time.ctime())print('本次扫描共检测到本网络存在%s台设备' % live_ip)搜索ip然后连接
admin@192.168.0.103's password:┌──────────────────────────────────────────────────────────────────────┐│ • MobaXterm Personal Edition v24.2 • ││ (SSH client, X server and network tools) ││ ││ ⮞ SSH session to admin@192.168.0.103 ││ • Direct SSH : ✓ ││ • SSH compression : ✓ ││ • SSH-browser : ✓ ││ • X11-forwarding : ✓ (remote display is forwarded through SSH) ││ ││ ⮞ For more info, ctrl+click on help or visit our website. │└──────────────────────────────────────────────────────────────────────┘Welcome to Ubuntu 22.04.5 LTS (GNU/Linux 5.10.4-sophon-custom aarch64)* Documentation: https://help.ubuntu.com* Management: https://landscape.canonical.com* Support: https://ubuntu.com/pro
overlay / overlay rw,relatime,lowerdir=/media/root-ro,upperdir=/media/root-rw/overlay,workdir=/media/root-rw/overlay-workdir 0 0
/dev/mmcblk0p5 /media/root-rw ext4 rw,relatime 0 0
/dev/mmcblk0p4 /media/root-ro ext4 ro,relatime 0 0Last login: Sat Oct 25 19:58:59 2025 from 192.168.0.104
/usr/bin/xauth: file /home/admin/.Xauthority does not exist
To run a command as administrator (user "root"), use "sudo <command>".
See "man sudo_root" for details.admin@sophon:~$
查看cpu
admin@sophon:~$ lscpu
Architecture: aarch64CPU op-mode(s): 32-bit, 64-bitByte Order: Little Endian
CPU(s): 8On-line CPU(s) list: 0-7
Vendor ID: ARMModel name: Cortex-A53Model: 4Thread(s) per core: 1Core(s) per cluster: 4Socket(s): -Cluster(s): 2Stepping: r0p4BogoMIPS: 50.00
Vulnerabilities:Itlb multihit: Not affectedL1tf: Not affectedMds: Not affectedMeltdown: Not affectedSpec store bypass: Not affectedSpectre v1: Mitigation; __user pointer sanitizationSpectre v2: Not affectedSrbds: Not affectedTsx async abort: Not affected
输入bm-smi
在PC机使用docker镜像，构建环境开始转模型
ubuntu@ubuntu:~$ docker pull sophgo/tpuc_dev:latest
ubuntu@ubuntu:~$ docker run --privileged --name myname1234 -v /home/ubuntu/sxj:/workspace -it sophgo/tpuc_dev:latest
root@956cfa92dca0:/workspace# model_transform
2025/10/26 20:09:24 - INFO : TPU-MLIR v1.23-20250930
usage: model_transform.py [-h] --model_name MODEL_NAME --model_def MODEL_DEF [--model_extern MODEL_EXTERN] [--model_data MODEL_DATA] [--input_shapes INPUT_SHAPES] [--input_types INPUT_TYPES][--output_names OUTPUT_NAMES] [--test_input TEST_INPUT] [--test_result TEST_RESULT] [--cache_skip] [--tolerance TOLERANCE] [--excepts EXCEPTS][--add_postprocess {,yolov3,yolov5,yolov7,yolov8,yolov8_seg,yolov11,yolov11_seg,ssd,bnr,mmap2rgbmap}] [--pruning PRUNING] [--onnx_sim {,skip_fuse_bn}][--do_onnx_sim DO_ONNX_SIM] [--debug] [-V] --mlir MLIR [--patterns_count PATTERNS_COUNT] [--dynamic_shape_input_names DYNAMIC_SHAPE_INPUT_NAMES][--shape_influencing_input_names SHAPE_INFLUENCING_INPUT_NAMES] [--dynamic] [--path_yaml PATH_YAML] [--not_inference] [--op_custom_shape OP_CUSTOM_SHAPE] [--enable_maskrcnn][--replace_topk_indices REPLACE_TOPK_INDICES] [--yuv_type {YUV420_PLANAR,YUV_NV21,YUV_NV12}] [--use_rewriter_config] [--struct_optimize STRUCT_OPTIMIZE] [--log_level {0,1}][--resize_dims RESIZE_DIMS] [--keep_aspect_ratio] [--keep_ratio_mode {letterbox,short_side_scale}] [--mean [MEAN]] [--scale SCALE][--pixel_format {rgb,bgr,gray,rgba,gbrg,grbg,bggr,rggb}] [--channel_format {nhwc,nchw,none}] [--pad_value PAD_VALUE] [--pad_type {normal,center}][--preprocess_list PREPROCESS_LIST] [--debug_cmd DEBUG_CMD][unknown_params]
model_transform.py: error: the following arguments are required: --model_name, --model_def, --mlir转paddleocr的检测模型
root@956cfa92dca0:/workspace# model_transform --model_name PP_OCRv5_mobile_det --model_def PP_OCRv5_mobile_det.onnx --input_shapes [[1,3,640,640]] --mlir PP_OCRv5_mobile_det.mlirroot@956cfa92dca0:/workspace# model_deploy --mlir PP_OCRv5_mobile_det.mlir --quantize F16 --processor bm1684x --model PP_OCRv5_mobile_det_1684x_f16.bmodel二、安装python包 https://github.com/sophgo/sophon-sail