ESP32CAM_yolov5

Run yolov5 by ESP32CAM from WAN

Update Notes:

Passed updates

• Data missing

20220828

• Upload Files
• Update esp32_cam.py:
  • Optimize item name printing
  • Only print item which larger than 129600 (close to camera)
  • Only print item when result confidence >= 0.5

20220924

• Rename esp32_cam.py to yolo_esp32cam_recv.py
• Create client app code (ESP32CAM_send_no_display.py)
• Updates:
  • Now code will set up a server and recive connection data from client app
  • Change alert size to auto culculate by 1/4 of total volume
  • Took off all of codes about picture process in client app (opencv, PIL, etc.)
  • Made client app available on mobile device.

20221002

• Add playsound module to play audio files.
• Uploaded mp3 files of detect objects.
• Made program run with different threads for sending and receiving data.
• When run on Android, playsound module doesn't work, so switch to use sl4a for android app. (kept two version of programs)
• Still got a big bug to fix:
  • mediaPlay is not working:
  • When use parameter: play=True(which is default), the program stocked.
  • If not, the program keep running but no sounds was played.

20221002

• esp32 connection bug fix

20221123

• Added a simplified version for raspberry pi
• Use pygame to play sounds

20221125

• Update requirements file
• Update raspi version

20221202

• Add flask to requests
• Make it can be started by web request

20221208

• Issue solve - mkdir video automatically
• Set ESP32CAM a hostname, then find its IP by it.
• run.py - simply run this python code to start program with flask for waiting the request
• wsgi.py - Having selections for running on computer or raspi

20221210

• Issue solve - '.local' is requested on raspi when finding the IP by hostname

20221219

• Update README.md
• This project has frozen temporary. Contect me if you have any questions.

Notes:

Environments

• You need to run this program in Anaconda, setup the environment for yolov5
  • yolov5 only run in Anaconda environments, so I can't help you for that.

Hardware / Firmware

• ESP32S (For the Cane, nothing about this python code so far)
• ESP32-CAM
  • Arduino Code download: https://github.com/MingMingFish/AIoT_Window_to_the_Soul
  • Android App: https://drive.google.com/file/d/1n-ipW-DTSQYuS5Hc4o0e074lagmyemJ5/view?usp=share_link
• Setup Docs (Zh-TW): https://docs.google.com/document/d/1-WC2XtERjPyeNyG127vGYmuaendjVqWl/edit?usp=share_link&ouid=107249627742763639384&rtpof=true&sd=true

Missing YOLOv5 File?

• Downloaded yolov5 from:

  • https://github.com/ultralytics/yolov5

• Or change the code below:

  • model = torch.hub.load(repo_or_dir='yolov5',model='yolov5x',source='local') # s/m/l/x

  to

  • model = torch.hub.load('ultralytics/yolov5', 'yolov5s') # source = 'github' # as default

Object Models

• You can change the model by editing the model name, the program will download the model automatically.

  • yolov5n / yolov5s / yolov5m / yolov5l / yolov5x / etc.

  • Or check on official websites:

    • https://pytorch.org/hub/ultralytics_yolov5/