00:19:14 Kerrie Geil: Today's HPC reservation is under my name again. Type in the SLURM sbatch Args optional field: --reservation=kerrie.geil_86 00:19:52 Kerrie Geil: the website link: https://kerriegeil.github.io/NMSU-USDA-ARS-AI-Workshops/ 00:22:10 Kerrie Geil: Also FYI because a couple of people asked, the zoom link is always the same for all workshop days (including next week) 00:22:33 Elizabeth Chin: will these slides be available to us after the workshop? 00:22:48 Kerrie Geil: they are on the workshop website learn page 00:22:59 Elizabeth Chin: awesome thank you!! 00:54:15 Amy Hudson (she/her): ill look up but any bash tips for renaming folders? 00:54:24 Lucas Heintzman: How many *.mat files should be eliminated? 00:54:42 Laura Boucheron: mv Airplanes_Side_2 airplanes 00:54:52 Amy Hudson (she/her): great thx 00:55:17 Ren Ortega: Wait are we moving Airplanes_Side_2 to the airplanes folder in 101? Or are we renaming it? 00:55:24 Kerrie Geil: just renaming 00:55:30 Ren Ortega: Got it, thanks 00:55:30 Jennifer Woodward-Greene: are they check_progress, FeatureDetectionQuality and progress?? 00:55:44 Lucas Heintzman: Are the .mat files the "shortcuts"? 00:58:33 Amy Hudson (she/her): green check is in the participants tab 01:01:48 Jennifer Woodward-Greene: glob = short for global 01:11:33 Jennifer Woodward-Greene: Can you confirm, the annotation coordinates are done by human hand on the images as our ground truth? 01:12:12 Jennifer Woodward-Greene: Thanks 01:12:53 Elizabeth Chin: are both the bounding box and contour annotations required, or do they have different use cases? 01:14:35 Elizabeth Chin: thanks! 01:26:33 Kerrie Geil: break until 10:21 Mountain 01:26:34 Dylan Burruss: Can you paste it in the chat window? 01:27:38 Kerrie Geil: I'm working on getting the html version posted to the website so we don't have to junk up the chat with all the code 01:27:56 Dylan Burruss: Thanks! 01:37:34 Ren Ortega: I'm getting a syntax error I don't know what to do with, can someone help me before I move on? 01:50:09 Lucas Heintzman: Go gray scale images need to be transformed to HSV? 01:51:23 Elizabeth Chin: there is no gray2hsv (https://scikit-image.org/docs/dev/api/skimage.color.html) 01:57:26 Jennifer Woodward-Greene: citizen science, Google captchas! 02:00:47 Kerrie Geil: Ok the notebook with answers is now posted to the workshop learn page. Here's the direct link: https://kerriegeil.github.io/NMSU-USDA-ARS-AI-Workshops/tutorials/Tutorial2_Classical_Machine_Learning_complete.html 02:37:13 Lucas Heintzman: Is this a typo: 02:37:27 Lucas Heintzman: 'moments_hu" retuns "moments_hul" 02:55:21 Maria Laura Cangiano: f = np.array([props[0].area, props[0].bbox_area, props[0].convex_area,\ props[0].eccentricity, props[0].equivalent_diameter,\ props[0].euler_number, props[0].extent, props[0].filled_area,\ props[0].major_axis_length, props[0].minor_axis_length,\ props[0].perimeter, props[0].solidity]) hu = np.array(props[0].moments_hu) new_f = np.concatenate((f, hu), axis=None) 02:58:09 Kale Harbick: i think you can even eliminate the first line: 02:58:18 Kale Harbick: f = np.concatenate((f), axis=None) 03:07:50 Elizabeth Chin: is there any randomness associated with extract_texture_features? I’m getting different values for the emu image 03:08:01 Lucas Heintzman: Likewise 03:08:23 Jennifer Woodward-Greene: I have different values too 03:08:35 Amy Hudson (she/her): same 3.72965154e-01 for my first value 03:08:39 Elizabeth Chin: pretty different- first row is: 6.11324265e-01 5.59225027e-01 4.79792054e-01 4.29846997e-01 03:08:49 Jennifer Woodward-Greene: [1.56044800e+01 1.74694492e+01 2.01266035e+01 2.17261141e+01 1.57594074e+00 1.06371099e+00 5.02286894e-01 4.12576946e-01 03:09:30 Kerrie Geil: if you have run into a "FileNotFoundError" it is likely that you've got some extra files hanging out in the Annotations data folder. There should only be subdirectories under Annotations. There should be no extra files or shortcuts hanging out in there. Go up to the beginning of the notebook and go through the section where we cleaned up the data again to be sure you've deleted or renamed everything properly 03:11:50 Elizabeth Chin: I think the numbers are just shuffled. my ‘first row’ is your 5th or 6th row displayed. 03:13:48 Amy Hudson (she/her): nice, looks like my feature names are just shuffled too 03:14:01 Lucas Heintzman: Will you repeat the 4 functions that must be correct to proceed? 04:05:27 Kerrie Geil: for HPC users, there was a small error in all tutorial notebooks in the intro section about how to build the conda env on the hpc. I fixed it to match what we have on our "Before you come" page and reuploaded all the notebooks to the learn page. Just wanted to note here that if you re-download the ipynbs now that section will be slightly different than it was before. This update doesn't affect those running on laptops at all 04:20:05 Jennifer Woodward-Greene: So you don't bias the test 04:20:15 Maximilian Feldman: Need to build model on some data, then see how well model performs 04:22:11 ARS - Kossi Nouwakpo: How do we ensure that some classes are not overrepresented in the training vs testing 04:22:12 ARS - Kossi Nouwakpo: ? 04:24:40 ARS - Kossi Nouwakpo: Thank you 04:40:53 Lucas Heintzman: Can you elaborate on what process was used to determine if an image was applied to X_test vs. X_train? Was it randomly sample from the collection, or was it just the first entries? 04:41:43 Lucas Heintzman: OK, so this was a simple numerical assignment. Thanks 05:18:47 Jennifer Woodward-Greene: My output is only SVC(kernel='linear') 05:18:54 Lucas Heintzman: Likewise 05:18:58 Maria Laura Cangiano: same 05:19:03 Matthew.McEntire: same 05:19:06 Kevin Cole: same 05:19:32 Jennifer Woodward-Greene: SVC(kernel='linear', verbose=True) 05:19:35 Elizabeth Chin: [LibSVM] SVC(kernel='linear', verbose=True) 05:19:36 Jennifer Woodward-Greene: Not much! 05:19:39 Maria Laura Cangiano: not really 05:19:56 Jennifer Woodward-Greene: personal pc 05:20:01 Maria Laura Cangiano: I'm on my own machine 05:20:04 ARS - Kossi Nouwakpo: own machine 05:20:17 Amy Hudson (she/her): im on hpc and getting what jennifer is getting 05:21:44 Lucas Heintzman: Even your "full tutorial" results in SVC(kernel='linear') just FYI 05:22:03 Kerrie Geil: the static html notebook with outputs also does not have the verbose info printing. We'll make sure to get Laura's version of the notebook and post it