<div dir="ltr">Hello,<div><br></div><div>I've seen some incredible output from example datasets with OpenDroneMap.</div><div><br></div><div>Right now, I'm working with some researchers who fly a UAS with a GoPro Hero 3+ Silver camera to collect video. I'm trying to pull image frames from that video, remove the fisheye distortion, and run the images through OpenDroneMap.</div><div><br></div><div>I pull frames from the video with this command:</div><div><br></div><div>$ ffmpeg -i video.mp4 -r 1 %03d.jpg</div><div><br></div><div>Then, what I've been doing is removing barrel distortion (fish eye) from the images with $ imagemagick:</div><div><br></div><div>convert image.jpg -distort barrel '0.030530 -0.124312 -0.038543' corrected_image.jpg</div><div><br></div><div>I also add exif tags to the imagery using exiftool:</div><div><br></div><div><div>$ exiftool -make=Gopro corrected_image.jpg</div><div>$ exiftool -model=HD3 corrected_image.jpg</div><div>$ exiftool -FocalLength="2.8 mm" corrected_image.jpg</div><div><br></div><div>My challenge now is determining if 2.88 mm is still the correct exif tag for the image, having already successfully removed distortion. Also, I'm not sure how this will affect the sensor size in ccd_defs.json. I've tried running the images through OpenDroneMap both distorted and undistorted using the exif tags shown above, and each time I end up with an empty point cloud.</div><div><br></div><div>Matching key points appeared to go okay.</div><div>
<p><span><font size="1" face="monospace, monospace">   14151 matches (23.8%) took  6.63s for /home/ubuntu/images/reconstruction-with-image-size-2400/matches/0-1.txt<span>       </span>...........</font></span></p>
<p><span><font size="1" face="monospace, monospace">   14093 matches (23.7%) took  6.93s for /home/ubuntu/images/reconstruction-with-image-size-2400/matches/0-2.txt<span>       </span>.</font></span></p>
<p><span><font size="1" face="monospace, monospace">    1670 matches ( 2.8%) took  6.94s for /home/ubuntu/images/reconstruction-with-image-size-2400/matches/0-3.txt<span>       </span>...............................</font></span></p>
<p><span><font size="1" face="monospace, monospace">   17644 matches (28.4%) took  7.22s for /home/ubuntu/images/reconstruction-with-image-size-2400/matches/1-2.txt<span>       </span>.</font></span></p>
<p><span><font size="1" face="monospace, monospace">    1762 matches ( 2.8%) took  7.20s for /home/ubuntu/images/reconstruction-with-image-size-2400/matches/1-3.txt<span>       </span>.........................................</font></span></p>
<p><span><font size="1" face="monospace, monospace">    1901 matches ( 3.0%) took  7.18s for /home/ubuntu/images/reconstruction-with-image-size-2400/matches/2-3.txt<span></span></font></span></p><p><span></span></p><div style="color:rgb(34,34,34);font-family:arial,sans-serif;font-size:small;font-style:normal;font-variant:normal;font-weight:normal;letter-spacing:normal;line-height:normal;text-align:start;text-indent:0px;text-transform:none;white-space:normal;word-spacing:0px;background-color:rgb(255,255,255)"></div><p></p><div style="color:rgb(34,34,34);font-family:arial,sans-serif;font-size:small;font-style:normal;font-variant:normal;font-weight:normal;letter-spacing:normal;line-height:normal;text-align:start;text-indent:0px;text-transform:none;white-space:normal;word-spacing:0px;background-color:rgb(255,255,255)">But running pmvs seemed to maybe filter out all those key points? Running meshing returned an error that the point cloud was empty.</div><div style="color:rgb(34,34,34);font-family:arial,sans-serif;font-size:small;font-style:normal;font-variant:normal;font-weight:normal;letter-spacing:normal;line-height:normal;text-align:start;text-indent:0px;text-transform:none;white-space:normal;word-spacing:0px;background-color:rgb(255,255,255)"><br></div><div style="color:rgb(34,34,34);font-family:arial,sans-serif;font-size:small;font-style:normal;font-variant:normal;font-weight:normal;letter-spacing:normal;line-height:normal;text-align:start;text-indent:0px;text-transform:none;white-space:normal;word-spacing:0px;background-color:rgb(255,255,255)">I was wondering if anyone has a solution to processing Go Pro imagery or video or if anyone has some ideas about what focal length or CCD size would be appropriate for undistorted imagery.</div><div style="color:rgb(34,34,34);font-family:arial,sans-serif;font-size:small;font-style:normal;font-variant:normal;font-weight:normal;letter-spacing:normal;line-height:normal;text-align:start;text-indent:0px;text-transform:none;white-space:normal;word-spacing:0px;background-color:rgb(255,255,255)"><br></div><div style="color:rgb(34,34,34);font-family:arial,sans-serif;font-size:small;font-style:normal;font-variant:normal;font-weight:normal;letter-spacing:normal;line-height:normal;text-align:start;text-indent:0px;text-transform:none;white-space:normal;word-spacing:0px;background-color:rgb(255,255,255)">Thanks,</div><div style="color:rgb(34,34,34);font-family:arial,sans-serif;font-size:small;font-style:normal;font-variant:normal;font-weight:normal;letter-spacing:normal;line-height:normal;text-align:start;text-indent:0px;text-transform:none;white-space:normal;word-spacing:0px;background-color:rgb(255,255,255)">Aaron</div><p><span><font size="1" face="monospace, monospace"><br></font></span></p></div><div><b>Aaron Dennis</b><div><a href="mailto:aaron.p.dennis@gmail.com" target="_blank">aaron.p.dennis@gmail.com</a></div></div>
</div></div>