Visually observe the leaves of different labels,Diseased leaves may be yellowish or brown.When change the brights or hsv,healthy leaves looks like diseased leaves.
\nHas anyone verified hsv enhance really usefull?
Visually observe the leaves of different labels,Diseased leaves may be yellowish or brown.When change the brights or hsv,healthy leaves looks like diseased leaves.
\nHas anyone verified hsv enhance really usefull?
For me, this technique worked. Keep in mind that any augmentation technique is having a lot of parameters. They have settings in which you can tune their specific intensity level and other specific parameters. My advice is to try to play will the augmentation parameters on a representative set of images containing all labels and find an optimum set of parameters
", "rawMarkdown": "For me, this technique worked. Keep in mind that any augmentation technique is having a lot of parameters. They have settings in which you can tune their specific intensity level and other specific parameters. My advice is to try to play will the augmentation parameters on a representative set of images containing all labels and find an optimum set of parameters", "votes": null }, { "id": "1119984", "postDate": "12/20/2020 14:10:03", "content": "@vladvdv thank you so much,I‘ll try to adjust these parameters
", "rawMarkdown": "vladvdv thank you so much,I‘ll try to adjust these parameters", "votes": null }, { "id": "1119986", "postDate": "12/20/2020 14:12:45", "content": "It is my pleasure. Good luck in your experiments !
", "rawMarkdown": "It is my pleasure. Good luck in your experiments !", "votes": null }, { "id": "1186695", "postDate": "02/05/2021 02:17:13", "content": "it is perplexing, indeed, now that you mention it - that modifying the hsv values would alter what we think is the precise distinguishing feature (color) between classes - but in a notebook that i forked, it did use an hsv shift augmentation. and when i removed it, it scored lower. my takeaway is that there is something to be gained by using that augmentation in the dataloader.
\nhere is that piece:
A.HueSaturationValue(\n hue_shift_limit=0.2, \n sat_shift_limit=0.2, \n val_shift_limit=0.2, \n p=0.5\n ),\nA is albumentation
", "rawMarkdown": "it is perplexing, indeed, now that you mention it - that modifying the hsv values would alter what we think is the precise distinguishing feature (color) between classes - but in a notebook that i forked, it did use an hsv shift augmentation. and when i removed it, it scored lower. my takeaway is that there is something to be gained by using that augmentation in the dataloader.\nhere is that piece:\n```\nA.HueSaturationValue(\n hue_shift_limit=0.2, \n sat_shift_limit=0.2, \n val_shift_limit=0.2, \n p=0.5\n ),\n```\nA is albumentation", "votes": null }, { "id": "1187537", "postDate": "02/05/2021 14:17:20", "content": "@daveccampbell thanks for your kind reply,now i'am using this in my code
", "rawMarkdown": "daveccampbell thanks for your kind reply,now i'am using this in my code", "votes": null }, { "id": "1198235", "postDate": "02/12/2021 20:03:59", "content": "Using a narrow range of these augmentations worked for me
", "rawMarkdown": "Using a narrow range of these augmentations worked for me", "votes": null } ], "comments": [ { "id": 1119852, "author_name": "vladvdv", "author_url": "", "post_date": "12/20/2020 12:01:37", "content": "For me, this technique worked. Keep in mind that any augmentation technique is having a lot of parameters. They have settings in which you can tune their specific intensity level and other specific parameters. My advice is to try to play will the augmentation parameters on a representative set of images containing all labels and find an optimum set of parameters
", "votes": null, "replies": [ { "id": 1119984, "author_name": "fatliuyun", "author_url": "", "post_date": "12/20/2020 14:10:03", "content": "@vladvdv thank you so much,I‘ll try to adjust these parameters
", "votes": null, "replies": [] }, { "id": 1119986, "author_name": "vladvdv", "author_url": "", "post_date": "12/20/2020 14:12:45", "content": "It is my pleasure. Good luck in your experiments !
", "votes": null, "replies": [] } ] }, { "id": 1186695, "author_name": "daveccampbell", "author_url": "", "post_date": "02/05/2021 02:17:13", "content": "it is perplexing, indeed, now that you mention it - that modifying the hsv values would alter what we think is the precise distinguishing feature (color) between classes - but in a notebook that i forked, it did use an hsv shift augmentation. and when i removed it, it scored lower. my takeaway is that there is something to be gained by using that augmentation in the dataloader.
\nhere is that piece:
A.HueSaturationValue(\n hue_shift_limit=0.2, \n sat_shift_limit=0.2, \n val_shift_limit=0.2, \n p=0.5\n ),\nA is albumentation
", "votes": null, "replies": [ { "id": 1187537, "author_name": "fatliuyun", "author_url": "", "post_date": "02/05/2021 14:17:20", "content": "@daveccampbell thanks for your kind reply,now i'am using this in my code
", "votes": null, "replies": [] } ] }, { "id": 1198235, "author_name": "vickygoyal", "author_url": "", "post_date": "02/12/2021 20:03:59", "content": "Using a narrow range of these augmentations worked for me
", "votes": null, "replies": [] } ], "raw_markdown_by_id": { "1119398": "Visually observe the leaves of different labels,Diseased leaves may be yellowish or brown.When change the brights or hsv,healthy leaves looks like diseased leaves.\nHas anyone verified hsv enhance really usefull?", "1119852": "For me, this technique worked. Keep in mind that any augmentation technique is having a lot of parameters. They have settings in which you can tune their specific intensity level and other specific parameters. My advice is to try to play will the augmentation parameters on a representative set of images containing all labels and find an optimum set of parameters", "1119984": "vladvdv thank you so much,I‘ll try to adjust these parameters", "1119986": "It is my pleasure. Good luck in your experiments !", "1186695": "it is perplexing, indeed, now that you mention it - that modifying the hsv values would alter what we think is the precise distinguishing feature (color) between classes - but in a notebook that i forked, it did use an hsv shift augmentation. and when i removed it, it scored lower. my takeaway is that there is something to be gained by using that augmentation in the dataloader.\nhere is that piece:\n```\nA.HueSaturationValue(\n hue_shift_limit=0.2, \n sat_shift_limit=0.2, \n val_shift_limit=0.2, \n p=0.5\n ),\n```\nA is albumentation", "1187537": "daveccampbell thanks for your kind reply,now i'am using this in my code", "1198235": "Using a narrow range of these augmentations worked for me" }, "source": "meta" }