multi-label-image-classification
multi-label-image-classification copied to clipboard
hellowangqian
Data augmentation M1,M2,M3 in the paper
Hello, I'm very interested in your implementation and I want to implement your code to get a start. However, I was wondering the meaning of M1, M2, M3 in your paper. I mean, does it mean: M1: randomflipping M2: randomresizecrop M3: mixup or M1: randomflipping M2: randomfilpping + randomresizecrop M3: randomfilpping + randomresizecrop + mixup
I'm looking forward to your answer, thanks in advance!
The later is what I mean, thank you for pointing out the confusion.
Get Outlook for Androidhttps://aka.ms/ghei36
From: kevin655 [email protected] Sent: Tuesday, December 8, 2020 6:25:32 AM To: hellowangqian/multi-label-image-classification [email protected] Cc: Subscribed [email protected] Subject: [hellowangqian/multi-label-image-classification] Data augmentation M1,M2,M3 in the paper (#3)
Hello, I'm very interested in your implementation and I want to implement your code to get a start. However, I was wondering the meaning of M1, M2, M3 in your paper. I mean, does it mean: M1: randomflipping M2: randomresizecrop M3: mixup or M1: randomflipping M2: randomfilpping + randomresizecrop M3: randomfilpping + randomresizecrop + mixup
I'm looking forward to your answer, thanks in advance!
— You are receiving this because you are subscribed to this thread. Reply to this email directly, view it on GitHubhttps://github.com/hellowangqian/multi-label-image-classification/issues/3, or unsubscribehttps://github.com/notifications/unsubscribe-auth/ABX467VAP3T6WUJMSZI3CH3STXBFZANCNFSM4URQBASA.
The later is what I mean, thank you for pointing out the confusion. Get Outlook for Androidhttps://aka.ms/ghei36 … ________________________________ From: kevin655 [email protected] Sent: Tuesday, December 8, 2020 6:25:32 AM To: hellowangqian/multi-label-image-classification [email protected] Cc: Subscribed [email protected] Subject: [hellowangqian/multi-label-image-classification] Data augmentation M1,M2,M3 in the paper (#3) Hello, I'm very interested in your implementation and I want to implement your code to get a start. However, I was wondering the meaning of M1, M2, M3 in your paper. I mean, does it mean: M1: randomflipping M2: randomresizecrop M3: mixup or M1: randomflipping M2: randomfilpping + randomresizecrop M3: randomfilpping + randomresizecrop + mixup I'm looking forward to your answer, thanks in advance! — You are receiving this because you are subscribed to this thread. Reply to this email directly, view it on GitHub<#3>, or unsubscribehttps://github.com/notifications/unsubscribe-auth/ABX467VAP3T6WUJMSZI3CH3STXBFZANCNFSM4URQBASA.
Thank you for your accurately reply, one thing I also want to know is that when using mixup in COCO, it's better to train more steps before decay learining rate ? (I saw that in your 'resnet101_model1fc.py', the step_size was set 'step_size=5'). Best wishes.