PROBLEM WITH RESOLUTION IMAGES 1944x27136 pixels (30,596 KB)

[marcosmoraisjr]

Hello everyone!

We are working with research on image annotations at the university (UEFS/BR).

In COCO Annotation, CVAT Annotation and LabelMe Annotation, we can upload and annotate images up to 1910x8815 px (20,322 KB).

However, we are unable to load high resolution images, for example: 1944x27136 px (30,596 KB).

Can you tell us something about it? Today, what would be the tool's resolution limit?

We appreciate your cooperation!

Marcos Morais Computer Science master’s degree, Information Technology post-graduate and Bachelor in Information Systems linktr.ee/marcosmoraisjr

Teaching is the highest form of understanding. Aristotle, Greek philosopher (384 - 322 BC)

[vpeopleonatank]

Can you share your example images? Did you try Via annotation tool - a lightweight annotator with plain javascript.

[marcosmoraisjr]

ds_teste2.rar https://drive.google.com/file/d/1I54WiRKEM_NWjo-bldI-bGLf4GcmsS2D/view?usp=drive_web ds_teste.rar https://drive.google.com/file/d/1c_s0kKFU6OLJlQqQ00iOS1F6_ptvjWJJ/view?usp=drive_web Good Morning! Here are some images I'm testing... You will see that some images do not load.

Thanks

Marcos Morais Computer Science master’s degree, Information Technology post-graduate and Bachelor in Information Systems linktr.ee/marcosmoraisjr https://l.instagram.com/?u=https%3A%2F%2Flinktr.ee%2Fmarcosmoraisjr&e=ATPYboqBwoVQHLcktc4wRaBKcLwRRTuDyRjQqpSuX0A8VAkapu1lEwChNT_jKdPAqc7b_eQicLlcPr_eUR7m4Q&s=1

Teaching is the highest form of understanding. Aristotle, Greek philosopher (384 - 322 BC)

Em seg., 19 de jul. de 2021 às 05:54, vpeopleonatank < @.***> escreveu:

Can you share your example images? Did you try Via annotation tool https://gitlab.com/vgg/via - a lightweight annotator with plain javascript.

— You are receiving this because you authored the thread. Reply to this email directly, view it on GitHub https://github.com/jsbroks/coco-annotator/issues/485#issuecomment-882371398, or unsubscribe https://github.com/notifications/unsubscribe-auth/AGNYOO4L5RBTESKKNI45NHTTYPR3VANCNFSM5AQFAIOA .

[vpeopleonatank]

I tried some web-based annotattion tools, It also doesn't work with large images above 20k pixel in each width and height dimension. But If I use labelme - a qt-based annotation tool. I can load and label images with an acceptable performance. CVAT annotator has proposed some approach to load large images (tiling image), but currently not implemented.

[marcosmoraisjr]

Helo! I'm doing research in the field of medical image annotations in a collaborative way. These images appreciate the high resolutions that the work of nephropathologists develops. Tell me a little more about your experience with WEB tools and LabelMe.

Ok, thanks.

Marcos Morais Computer Science master’s degree, Information Technology post-graduate and Bachelor in Information Systems linktr.ee/marcosmoraisjr https://l.instagram.com/?u=https%3A%2F%2Flinktr.ee%2Fmarcosmoraisjr&e=ATPYboqBwoVQHLcktc4wRaBKcLwRRTuDyRjQqpSuX0A8VAkapu1lEwChNT_jKdPAqc7b_eQicLlcPr_eUR7m4Q&s=1

Teaching is the highest form of understanding. Aristotle, Greek philosopher (384 - 322 BC)

Em seg., 19 de jul. de 2021 às 22:53, vpeopleonatank < @.***> escreveu:

I tried some web-based annotattion tools, It also doesn't work with large images above 20k pixel in each width and height dimension. But If I use labelme - a qt-based annotation tool. I load and label images with an acceptable performance. CVAT annotator has proposed some approach to load large images (tiling image), but currently not implemented.

— You are receiving this because you authored the thread. Reply to this email directly, view it on GitHub https://github.com/jsbroks/coco-annotator/issues/485#issuecomment-882985826, or unsubscribe https://github.com/notifications/unsubscribe-auth/AGNYOO5U7BXIPROQUDQ4CB3TYTJKPANCNFSM5AQFAIOA .

[vpeopleonatank]

I 've just installed another labeling tool for medical images (read dicom format) https://github.com/vinbigdata-medical/vindr-lab. I setup completely by walking through docker installation guide https://github.com/vinbigdata-medical/vindr-lab-deployment. But I don't have dicom file, I tried to convert your jpg images to dicom format (.dcm extension) and upload it to vindr-lab but I get parsing error message. You can try it on your own, but if you can share some right large dicom fomat images, I also can try it too .

[marcosmoraisjr]

ds_teste7.rar https://drive.google.com/file/d/11hhLIzrKnnB8DvkL3ZYUhvXBMFjavHnq/view?usp=drive_web Hello everyone!

Using as selection criteria, images (PNG or JPG) that are not loaded in the COCO Annotator and CVAT annotation software, an image with 30000x20000 px (42,472 KB) was chosen which served as a matrix to generate new images (ds_teste7) with different data characteristics and submitted to the mentioned annotation software, resulting in the table below:

Attached I am sending you a set of images. If you can't, let me know that I have a drive available on the internet for you to download.

Marcos Morais

*Computer Science master’s degree, Information Technology post-graduate andBachelor in Information Systems*linktr.ee/marcosmoraisjr https://l.instagram.com/?u=https%3A%2F%2Flinktr.ee%2Fmarcosmoraisjr&e=ATPYboqBwoVQHLcktc4wRaBKcLwRRTuDyRjQqpSuX0A8VAkapu1lEwChNT_jKdPAqc7b_eQicLlcPr_eUR7m4Q&s=1

Teaching is the highest form of understanding. Aristotle, Greek philosopher (384 - 322 BC)

Em ter., 20 de jul. de 2021 às 10:54, vpeopleonatank < @.***> escreveu:

I 've installed another labeling tool for medical images (read dicom format) https://github.com/vinbigdata-medical/vindr-lab. I setup completely by walking through docker installation guide https://github.com/vinbigdata-medical/vindr-lab-deployment. But I don't have dicom file, I tried to convert your jpg images to dicom format (.dcm extension) and upload it to vindr-lab but I get parsing error message. You can try it on your own, but if you can share some right large dicom fomat images, I also can try it too .

— You are receiving this because you authored the thread. Reply to this email directly, view it on GitHub https://github.com/jsbroks/coco-annotator/issues/485#issuecomment-883413269, or unsubscribe https://github.com/notifications/unsubscribe-auth/AGNYOO6A62HWKLPLP6DBEFTTYV52TANCNFSM5AQFAIOA .

[vpeopleonatank]

I 've tried to open 1944x27136 resolution image, but vindr-lab returns Out of memory at ImageData creation. Currently, I don't know what labeling tool indeed supporting large images. I think you should contribute to implement tiling feature in cvat annotator or temporarily using desktop label tool like Labelme.

[marcosmoraisjr]

Thanks! If you need something, please contact me partner!

Marcos Morais linktr.ee/marcosmoraisjr https://l.instagram.com/?u=https%3A%2F%2Flinktr.ee%2Fmarcosmoraisjr&e=ATPYboqBwoVQHLcktc4wRaBKcLwRRTuDyRjQqpSuX0A8VAkapu1lEwChNT_jKdPAqc7b_eQicLlcPr_eUR7m4Q&s=1

Teaching is the highest form of understanding.Aristotle, Greek philosopher (384 - 322 BC)

Em qua., 21 de jul. de 2021 às 21:41, vpeopleonatank < @.***> escreveu:

I 've tried to open 1944x27136 resolution image, but vindr-lab returns Out of memory at ImageData creation. Currently, I don't know what labeling tool indeed supporting large images. I think you should contribute to implement tiling feature in cvat annotator or temporarily using desktop label tool like Labelme.

— You are receiving this because you authored the thread. Reply to this email directly, view it on GitHub https://github.com/jsbroks/coco-annotator/issues/485#issuecomment-884586893, or unsubscribe https://github.com/notifications/unsubscribe-auth/AGNYOO676C7VFQNWLHC7QSDTY5SNLANCNFSM5AQFAIOA .

[DenysIsaiev]

There are a problem during scan dataset: pythons PIL.Image has maximum pixels. It can be fixed by adding Image.MAX_IMAGE_PIXELS = None in database.images file. Unfortunately after adding this image to database there are another problem: big images loads too slow and paperjs throws timeout