KyanChen/RSPrompter issues and pull requests

#100 - 请问，有人训练成功了吗？

Issue - State: open - Opened by AImagicAI 10 months ago - 13 comments

#99 - 在使用demo.py中自带图片进行预测时，输出文件夹无法访问，或者缺少参数

Issue - State: open - Opened by hailongm 10 months ago

#98 - loss=0

Issue - State: open - Opened by AImagicAI 10 months ago

#97 - 目前用建筑数据集和自己的数据集生成的训练json文件，模型可以跑起来，但是bboxloss一直为零，用自带的nwpu的训练集，bboxloss不为零，模型能正常收敛，有老师遇到过吗

Issue - State: open - Opened by funny000 10 months ago - 5 comments

#96 - no pos rois

Issue - State: closed - Opened by AImagicAI 10 months ago - 8 comments

#95 - 我的损失一开始收敛，但是很快就变成NaN了，我想知道这可能是什么原因

Issue - State: open - Opened by 21-10-4 11 months ago - 1 comment

#94 - 请问当用一个本地数据集微调的时候，模型的泛化性是不是会降低？

Issue - State: open - Opened by funny000 11 months ago

#93 - 如何做分割

Issue - State: closed - Opened by l-cr 11 months ago

#92 - no pos rois

Issue - State: closed - Opened by xyLpf about 1 year ago - 6 comments

#91 - 关于LoRA高效微调方法使得输入图像尺寸可变的问题

Issue - State: open - Opened by sleepat-11 about 1 year ago - 1 comment

#90 - Are any pretrained model weights provided?

Issue - State: closed - Opened by learner-NKH about 1 year ago - 1 comment

#89 - 使用image_demo图像预测的问题

Issue - State: open - Opened by cricket-111 about 1 year ago - 6 comments

#88 - 怎么没找到将DeepSpeed训练的模型转换为MMDetection的模型的代码 ‘’zero_to_fp32.py‘’在哪啊

Issue - State: closed - Opened by moderate-leon about 1 year ago - 1 comment

#87 - 有同学尝试将训练的模型转为onnx的吗

Issue - State: closed - Opened by Jinunmeng about 1 year ago - 5 comments

#86 - lightning版的代码中用prompter-anchor进行训练验证时指标都没问题（AP60多），训练完成后用tools/predict.py和config里面的predict.py代码进行测试时，输出的结果一直是空（图像上面没有预测框和分割结果）。同样是在Vit-huge中加入一些可学习参数后，lightning版的代码训练速度没有慢多少，新版mmdet代码训练速度非常慢（在3090上nwpu数据集50个iter要40分钟，但用训练得到的权重测试的时候速度是正常的），这两个问题困扰我好长时间了，有大佬有代码在lightning版可以正常出预测结果吗？

Issue - State: open - Opened by moderate-leon about 1 year ago - 6 comments

#85 - File download error-- not pytorch_model.bin but model.safetensors

Issue - State: open - Opened by qsd-github about 1 year ago - 2 comments

#84 - 显存占用和模型准确度的问题

Issue - State: closed - Opened by stivory about 1 year ago - 1 comment

#83 - 遥感大幅影像的实例拼接

Issue - State: open - Opened by GuoxingYan about 1 year ago - 1 comment

#82 - 请问RandomResize如何缩放实例分割标注的？

Issue - State: closed - Opened by DotWang about 1 year ago - 2 comments

#81 - nwpu数据集map问题

Issue - State: closed - Opened by wzr0108 about 1 year ago - 3 comments

#80 - The model and loaded state dict do not match exactly

Issue - State: closed - Opened by Yisng about 1 year ago - 4 comments

#79 - 作者您好，请问configs/rsprompter/samseg-maskrcnn-nwpu.py有freeze RSSamVisionEncoder吗？

Issue - State: closed - Opened by wzr0108 about 1 year ago - 1 comment

#78 - I have been unable to connect to the huggingface URL. Is there any other way without linking to the huggingface URL?

Issue - State: closed - Opened by moderate-leon about 1 year ago - 2 comments

#77 - ImportError: Failed to import custom modules from {'imports': ['mmdet.rsprompter'], 'allow_failed_imports': False}, the current sys.path is:

Issue - State: open - Opened by Yisng about 1 year ago - 5 comments

#76 - 一张3090能跑RSPrompter-anchor吗

Issue - State: closed - Opened by wzr0108 about 1 year ago - 1 comment

#75 - Does it support datasets in VOC format？

Issue - State: closed - Opened by Stark320 about 1 year ago - 1 comment

#74 - inference on custom dataset

Issue - State: closed - Opened by 15926273249 about 1 year ago - 4 comments

#73 - Compare the time spent for each epoch between the latest version and the original one

Issue - State: closed - Opened by pkyzh2006 about 1 year ago - 5 comments

#72 - Cannot find ‘RSPrompterQuery’

Issue - State: closed - Opened by pkyzh2006 about 1 year ago - 1 comment

#71 - what is mmpl?

Issue - State: closed - Opened by xypu98 about 1 year ago - 1 comment

#70 - 全论文我看到的只有框提示自动生成，并没有点提示自动生成，请问点提示自动生成可以实现吗？？

Issue - State: closed - Opened by Daisygotit666 about 1 year ago - 1 comment

#69 - query-based方法中学习出的4*256的prompt embedding包含的是位置信息还是image embedding信息？

Issue - State: closed - Opened by sunsunnyshine about 1 year ago - 1 comment

#68 - 想问为什么一直没安装好包就反反复复一直提示什么什么不行但是到torchvision再也修复不了了

Issue - State: closed - Opened by Sophia-924 about 1 year ago - 1 comment

#67 - 能否上传关于SAM_cls的配置文件呢

Issue - State: closed - Opened by KeyCatcc about 1 year ago - 1 comment

#66 - RSPrompter-anchor

Issue - State: closed - Opened by cyq123-git over 1 year ago - 1 comment

#65 - 为什么同样使用同一个虚拟环境，我一直报这个错误，IndexError: list index out of range，求解，谢谢

Issue - State: closed - Opened by gangtieyang over 1 year ago - 5 comments

#64 - size dismatch error when using weights NWPU_anchor.pth

Issue - State: closed - Opened by Kevinorve over 1 year ago - 2 comments

#63 - Why do you need to train so many rounds?

Issue - State: closed - Opened by Nastu-Ho over 1 year ago - 5 comments

#62 - KeyError: 'pytorch-lightning_version'

Issue - State: closed - Opened by bugcat9 over 1 year ago - 4 comments

#61 - 请问可以提供一下RSPrompter-query的权重文件吗

Issue - State: closed - Opened by cricket-111 over 1 year ago - 1 comment

#60 - from instance seg to semantic seg

Issue - State: closed - Opened by Nastu-Ho over 1 year ago - 2 comments

#59 - pretrain/sam/sam_vit_h_4b8939.pth在哪里啊

Issue - State: closed - Opened by kungkook over 1 year ago - 1 comment

#58 - Does it support multiple bands, such as Landsat and Sentinel S2?

Issue - State: closed - Opened by 17701311118 over 1 year ago - 1 comment

#57 - 使用NWPU数据集，训练过程中loss值正常，验证时精度一直为0。

Issue - State: closed - Opened by moderate-leon over 1 year ago - 6 comments

#56 - How to change the input image size?

Issue - State: closed - Opened by pengyuting181 over 1 year ago - 2 comments

#55 - Does mask_decoder in RSPrompter participate in training?

Issue - State: closed - Opened by willpat1213 over 1 year ago - 1 comment

#54 - Question about query-based prompter.

Issue - State: closed - Opened by willpat1213 over 1 year ago - 1 comment

#53 - 你们训练一个epoch大概多久，2080ti训练了一天才8个epoch 合理吗

Issue - State: closed - Opened by jxhoh over 1 year ago - 2 comments

#52 - 做预测的时候。。怎么设置预测的保存路径啊

Issue - State: closed - Opened by jxhoh over 1 year ago - 1 comment

#51 - 你们都是怎么做预测的

Issue - State: closed - Opened by jxhoh over 1 year ago - 8 comments

#50 - Is there anyone who can get the performance in the paper? I can’t get the performance no matter how I adjust the config on nwpu.

Issue - State: closed - Opened by YananGu over 1 year ago - 2 comments

#49 - 你们的pth文件都保存在哪里？？？

Issue - State: closed - Opened by jxhoh over 1 year ago - 3 comments

#48 - Anyone encountered this KeyError: 'pytorch-lightning_version'? It seems come from the provided checkpoint.

Issue - State: closed - Opened by qixycn over 1 year ago - 6 comments

#47 - Predict.py file running error

Issue - State: closed - Opened by pkyzh2006 over 1 year ago - 1 comment

#46 - 阅读论文关于提示生成的疑问？

Issue - State: closed - Opened by ZHIZIHUABU over 1 year ago - 1 comment

#45 - Question about the performance

Issue - State: closed - Opened by YananGu over 1 year ago - 6 comments

#44 - query-base训练不收敛

Issue - State: closed - Opened by an1018 over 1 year ago - 5 comments

#43 - My training data consists of RGB images and binary black-and-white masks (black: background, white: objects). How should I adjust the model?

Issue - State: closed - Opened by pkyzh2006 over 1 year ago

#42 - On other datasets

Issue - State: closed - Opened by YananGu over 1 year ago - 1 comment

#41 - Could you provide your ckpt file conveniently?

Issue - State: closed - Opened by Yanmoche over 1 year ago - 2 comments

#40 - backbone使用sam_vit_b和sam_vit_h效果差不多

Issue - State: closed - Opened by LIUTAOGE over 1 year ago - 2 comments

#39 - type Error: File "/usr/local/lib/python3.10/dist-packages/lightning/fabric/utilities/cloud_io.py", line 83, in _is_object_storage from adlfs import GCSFileSystem

Issue - State: closed - Opened by mamad315 over 1 year ago - 1 comment

#38 - 数据集标签转换

Issue - State: closed - Opened by WanFeng123313 over 1 year ago - 3 comments

#37 - 能否透露预计整合进mmdet的时间？

Issue - State: closed - Opened by Xie-Muxi over 1 year ago - 1 comment

#36 - 实验配置

Issue - State: closed - Opened by lebron-2016 over 1 year ago - 1 comment

#35 - How to fine-tune the decoder of sam?

Issue - State: closed - Opened by willpat1213 over 1 year ago - 4 comments

#34 - train.py出现FileNotFoundError

Issue - State: closed - Opened by winglucky over 1 year ago - 1 comment

#33 - How should I use trained weights to infer my own images

Issue - State: closed - Opened by HGD-ai over 1 year ago - 1 comment

#32 - Why are my results all 0 during training?

Issue - State: closed - Opened by Fairywyt over 1 year ago - 3 comments

#31 - The result is not the same with paper

Issue - State: closed - Opened by wuwuwuxxx over 1 year ago - 3 comments

#30 - No `test_step()` method defined to run `Trainer.test`

Issue - State: closed - Opened by LIUTAOGE over 1 year ago - 1 comment

#29 - A test environment problem occurs KeyError: 'pytorch-lightning_version'

Issue - State: closed - Opened by liuhongyan123456 over 1 year ago - 1 comment

#28 - size mismatch for panoptic_head.roi_head.mask_head.point_emb.8.weight:

Issue - State: closed - Opened by jainie-max over 1 year ago

#27 - when it will be available on the mmdectetion?

Issue - State: closed - Opened by Jasper0420 over 1 year ago - 2 comments

#26 - SAM-cls

Issue - State: closed - Opened by gykj-hmf over 1 year ago - 1 comment

#25 - predict.py报错

Issue - State: closed - Opened by LIUTAOGE over 1 year ago - 1 comment

#24 - How to resume form ckpt?

Issue - State: closed - Opened by Z-MU-Z over 1 year ago - 2 comments

#23 - How to increase mIoU evaluation indicators?

Issue - State: closed - Opened by Xie-Muxi over 1 year ago - 1 comment

#22 - about Visualization

Issue - State: closed - Opened by YananGu over 1 year ago - 5 comments

#21 - How to train well on single GPU?

Issue - State: closed - Opened by sandeepjana over 1 year ago - 1 comment

#20 - Cannot load ".pth" trained model file provided in this code.

Issue - State: closed - Opened by zfbi over 1 year ago - 3 comments

#19 - HuggingFace demo shows runtime error

Issue - State: closed - Opened by HumbleGamer over 1 year ago - 1 comment

#18 - How to save checkpoints and visualize during training?

Issue - State: closed - Opened by JaireYu over 1 year ago - 3 comments

#17 - Where is the ”sam_vit_h_4b8939.pth file“

Issue - State: closed - Opened by Jasper0420 over 1 year ago - 2 comments

#16 - TypeError: 'list' object cannot be interpreted as an integer,when running the predict.py

Issue - State: closed - Opened by big-boss26 over 1 year ago - 9 comments

#15 - Verify the result visualization

Issue - State: closed - Opened by ZhaoPo over 1 year ago - 2 comments

#14 - Unable to download datasets

Issue - State: closed - Opened by sandeepjana over 1 year ago - 2 comments

#13 - KeyError: 'pytorch-lightning_version'

Issue - State: closed - Opened by dongwhfdyer over 1 year ago - 1 comment

GitHub / KyanChen/RSPrompter issues and pull requests