User Tools

Site Tools


journal

Differences

This shows you the differences between two versions of the page.

Link to this comparison view

Next revision
Previous revision
journal [2022/09/23 02:24] – created jhagstrandjournal [2023/01/12 11:51] (current) – removed jhagstrand
Line 1: Line 1:
-====== Journal ====== 
- 
-13 Jun<br/> 
-Purchase mini traffic cones from lazada.  Red. 
- 
-29 Jun<br/> 
-Take photos of a red traffic cones from different angles. Using Android phone. 
- 
-30 Jun<br/> 
-Upload photos to Google Drive folder<br/>====== Journal ====== 
- 
-13 Jun \\ 
-Purchase mini traffic cones from lazada.  Red. 
- 
-29 Jun \\ 
-Take photos of a red traffic cones from different angles. Using Android phone. 
- 
-30 Jun \\ 
-Upload photos to Google Drive folder \\ 
-/My\ Drive/sync/skateboard/images/cones/raw 
- 
-1 Jul \\ 
-create skateboard.ipynb for use with Google Colab \\ 
-example https://medium.com/analytics-vidhya/train-yolo-v3-to-detect-custom-objects-with-free-gpu-d8d0262100fe 
- 
-2 Jul \\ 
-Image processing.  Scale down. \\ 
-$ python3 prepimage.py \\ 
-output into folder: \\ 
-/My\ Drive/sync/skateboard/images/cones/train 
- 
-Image labeling.  There are several labeling programs.  I used a free online service,  
-https://www.makesense.ai/ \\ 
-one file named classes.name \\ 
-one .txt file for each .jpg file \\ 
-output into same folder with the images: \\ 
-/My\ Drive/sync/skateboard/images/cones/train \\ 
-one row for each detected object \\ 
-5 fields per row: 
-  * index into label list in file classes.name, we use only one label: cone, index 0 
-  * x normalized as percentage of width 
-  * y (top to bottom) normalized as percentage of height 
-  * w normalized as percentage of width 
-  * h normalized as percentage of height 
- 
-13 Jul \\ 
-Received Tello drone. \\ 
-Had to do three things to get it to hover. 
-  - Calibrate. 
-  - Turn off the AC. 
-  - Lay a patterned sheet down on the floor. 
- 
-Tried two apps. 
-  - Tello.  official app from DJI/Ryze. 
-  - TelloMe.  from volate!lo.  Like Tello plus FollowMe capability.  volate!lo also makes TelloFPV, which I will not try. 
- 
-16 jul \\ 
-Flew Tello outside \\ 
-TelloMe does not do orbit and follow at the same time. \\ 
-Video drops frames with increased distanced. \\ 
- 
-31 Jul \\ 
-I've been running programs driving the Tello, testing sockets, and using OpenCV. \\ 
-Today, I got an error Failed to load module "canberra-gtk-module" \\ 
-  Traceback (most recent call last): 
-  File "color.py", line 118, in <module> 
-  img = cv2.imread(coneimg, CV_LOAD_IMAGE_COLOR); 
-  NameError: name 'CV_LOAD_IMAGE_COLOR' is not defined 
- 
-I ran  \\ 
-`sudo apt-get install python-opencv` \\ 
-and now `import cv2` says \\ 
-ModuleNotFoundError: No module named 'cv2' \\ 
- 
-`pip install opencv-python` says \\ 
-ERROR: Could not find a version that satisfies the requirement opencv-python (from versions: none) \\ 
-ERROR: No matching distribution found for opencv-python \\ 
- 
-Why would I use pip vs apt-get? \\ 
- 
-I ran  \\ 
-`sudo apt-get remove python-opencv` \\ 
-`sudo apt-get purge python-opencv` \\ 
-still no module named 'cv2' \\ 
- 
-Corrected using `sudo apt-get install python3-opencv` \\ 
-It installed version 3.2.  Don't know how to get 3.3.1. \\ 
- 
-/My\ Drive/sync/skateboard/images/cones/raw 
- 
-1 Jul<br/> 
-create skateboard.ipynb for use with Google Colab<br/> 
-example https://medium.com/analytics-vidhya/train-yolo-v3-to-detect-custom-objects-with-free-gpu-d8d0262100fe 
- 
-2 Jul<br/> 
-Image processing.  Scale down.<br/> 
-$ python3 prepimage.py<br/> 
-output into folder:<br/> 
-/My\ Drive/sync/skateboard/images/cones/train 
- 
-Image labeling.  There are several labeling programs.  I used a free online service,  
-https://www.makesense.ai/<br/> 
-one file named classes.name<br/> 
-one .txt file for each .jpg file<br/> 
-output into same folder with the images:<br/> 
-/My\ Drive/sync/skateboard/images/cones/train<br/> 
-one row for each detected object<br/> 
-5 fields per row: 
-* index into label list in file classes.name, we use only one label: cone, index 0 
-* x normalized as percentage of width 
-* y (top to bottom) normalized as percentage of height 
-* w normalized as percentage of width 
-* h normalized as percentage of height 
- 
-13 Jul 
-Received Tello drone. 
-Had to do three things to get it to hover. 
-1. Calibrate. 
-1. Turn off the AC. 
-1. Lay a patterned sheet down on the floor. 
- 
-Tried two apps. 
-1. Tello.  official app from DJI/Ryze. 
-1. TelloMe.  from volate!lo.  Like Tello plus FollowMe capability.  volate!lo also makes TelloFPV, which I will not try. 
- 
-16 jul 
-Flew Tello outside 
-TelloMe does not do orbit and follow at the same time. 
-Video drops frames with increased distanced. 
- 
-31 Jul 
-I've been running programs driving the Tello, testing sockets, and using OpenCV. 
-Today, I got an error Failed to load module "canberra-gtk-module" 
-Traceback (most recent call last): 
-  File "color.py", line 118, in <module> 
-    img = cv2.imread(coneimg, CV_LOAD_IMAGE_COLOR); 
-NameError: name 'CV_LOAD_IMAGE_COLOR' is not defined 
-I ran  
-`sudo apt-get install python-opencv` 
-and now `import cv2` says 
-ModuleNotFoundError: No module named 'cv2' 
- 
-`pip install opencv-python` says 
-ERROR: Could not find a version that satisfies the requirement opencv-python (from versions: none) 
-ERROR: No matching distribution found for opencv-python 
- 
-Why would I use pip vs apt-get? 
- 
-I ran  
-`sudo apt-get remove python-opencv` 
-`sudo apt-get purge python-opencv` 
-still no module named 'cv2' 
- 
-Corrected using `sudo apt-get install python3-opencv` 
-It installed version 3.2.  Don't know how to get 3.3.1. 
- 
- 
- 
- 
- 
- 
  
journal.1663914286.txt.gz · Last modified: 2022/09/23 02:24 by jhagstrand

Except where otherwise noted, content on this wiki is licensed under the following license: Public Domain
Public Domain Donate Powered by PHP Valid HTML5 Valid CSS Driven by DokuWiki