Audio4Vision #5 - Show and Tell: More Iterations and Finetuning

RE: Audio4Vision #5 - Show and Tell: More Iterations and Finetuning

aspork42 — Mon, 20 Aug 2018 03:32:41 GMT

Great update! I agree - there is a big difference between driving and being parked, but you have clearly explained what is happening and why that would be the case. Looking forward to seeing the project develop.

RE: Audio4Vision #5 - Show and Tell: More Iterations and Finetuning

genebren — Sun, 12 Aug 2018 16:29:32 GMT

Interesting update. I wonder on image 11.jpg, which caption is more correct (before or after). Before says 'driving', while after says 'parked'. There is a big difference between the two statements. Also, for motion, wouldn't a series of images (i.e. change in position or potential velocity) be a valuable input for a correct caption? Would your modeling allow for a caption based on differences between images?

Good luck,

Gene

RE: Audio4Vision #5 - Show and Tell: More Iterations and Finetuning

dixonselvan — Sun, 12 Aug 2018 15:29:04 GMT

Nice update [mention:d46d6caae03142c99cf969d112d3ab1c:e9ed411860ed4f2ba0265705b8793d05] But I don't seem to find any improvement after increasing the number of iterations to 2M (except the first one where it correctly identifies car is parked near a tree). In the second image case, increasing the iterations and fine-tuning has resulted in a wrong conclusion whereas 1M iterations have judged the scenario correctly.