sainivedh19pt

Results 10 comments of sainivedh19pt

Performance on Cifar10

Hi, I tried the same but val_acc is still below 25%...you had any success ?

Constant IoU after every validation

Hi @cspearl @MengzhangLI, Faced the same issue when trained using [latest codebase commit](https://github.com/open-mmlab/mmsegmentation/pull/2016) for binary class segmentation (roads class). mIOU, aACC, mACC is the same through out training. Attaching log...

[VATT] OP_REQUIRES failed at strided_slice_op.cc:108 : INVALID_ARGUMENT: slice index 0 of dimension 0 out of bounds.

Hi @hassanhub, Thanks for the wonderful work. I'm trying to reproduce the pretraining step of VATT but facing the same error as everyone above. ` OP_REQUIRES failed at strided_slice_op.cc:108 :...

How do I find your core code of GDL and PCB?

Hi @weixj1998, You can find the codebase at defrcn/evaluation/calibration_layer ----> PrototypicalCalibrationBlock

Missing category label

The labelling of classes in annotations.csv in totally wrong. Surprised by how none has raised it yet

[VATT] Construct and load datasets using DMVR

@G-Apple1 Hi, got the same error but I can surpass it when we use tf.funtion() decorator.

[VATT] Construct and load datasets using DMVR

@ybkim95 Sorry I may not be much of help as its been a long time

What is the difference between pedestrian and people, van and truck? They seem the one object.

Hi, Found this description given in the paper **If a human maintains standing pose or walking, we classify it as pedestrian; otherwise, it is classified as a person.** Thanks

[Bug] YOLOX "loss_bbox" may sometimes equal to zero in training.

Hi, I have observed that with **num_classes=20**, you have **use_sigmoid=True** in **loss_cls**. Should this be set to **False** for multi class training, otherwise it uses [binary_crossentropy loss](https://github.com/open-mmlab/mmdetection/blob/31c84958f54287a8be2b99cbf87a6dcf12e57753/mmdet/models/losses/cross_entropy_loss.py#L245). Thanks

Inconsistencies in the labels

Thanks for pointing this out, spent an hour figuiring out where I made the json parsing wrong !!.