Classifying Images of Clouds in the Cloud with AutoML Vision Reviews

Anderson M. · Reviewed أكثر من 3 سنوات ago

Dipsikha T. · Reviewed أكثر من 3 سنوات ago

Chandra K. · Reviewed أكثر من 3 سنوات ago

Cristian C. · Reviewed أكثر من 3 سنوات ago

Igor I. · Reviewed أكثر من 3 سنوات ago

Harindha F. · Reviewed أكثر من 3 سنوات ago

Joe W. · Reviewed أكثر من 3 سنوات ago

Aashis K. · Reviewed أكثر من 3 سنوات ago

great understandable presentation

chollangi c. · Reviewed أكثر من 3 سنوات ago

Riccardo S. · Reviewed أكثر من 3 سنوات ago

sangavi k. · Reviewed أكثر من 3 سنوات ago

Yi L. · Reviewed أكثر من 3 سنوات ago

Cheng L. · Reviewed أكثر من 3 سنوات ago

Yogesh J. · Reviewed أكثر من 3 سنوات ago

賢司 塩. · Reviewed أكثر من 3 سنوات ago

Laxman K. · Reviewed أكثر من 3 سنوات ago

Henry L. · Reviewed أكثر من 3 سنوات ago

Tanay D. · Reviewed أكثر من 3 سنوات ago

Pankaj K. · Reviewed أكثر من 3 سنوات ago

Leo B. · Reviewed أكثر من 3 سنوات ago

Aditi S. · Reviewed أكثر من 3 سنوات ago

Rahul J. · Reviewed أكثر من 3 سنوات ago

Definitely the worst lab so far. The instructions are out of date - e.g. CSV-based image import seems to require a header row, absent in the original data.csv; it succeeded only after I've added the labels; visiting web UI of GCS is confusing as it serves no purpose - CLI is used to copy all the files *and to list them*, so visit to UI is redundant and makes the student wonder if they've missed someting. I feel like I've learned nothing, it was just copying files from one bucket to another and clicking some buttons with default settings... Now for the constructive part: how about providing a pre-trained model to copy to student's account instead of telling them to "see the screenshots"? How about changing the dataset used for training (image sizes range from 50 KB to whopping 9 MB!) - if you're just showcasing the tool, maybe something smaller/simpler would suffice? I'm no expert in ML, but I suppose that teaching the model to recognize items on 64x64 px icons would be much faster. Or maybe provide 2 or 3 pre-trained models with different quality of input images to show the differences? Please review the lab and update the instructions, in its current form it's a waste of time and resources.

Andrzej G. · Reviewed أكثر من 3 سنوات ago

Vinayak H. · Reviewed أكثر من 3 سنوات ago

José T. · Reviewed أكثر من 3 سنوات ago