Thank you for your code it has been very helpful for me, however I'm encountering some issues regarding the object detection task:
If I do not insert the gradient clipping the loss will explode causing it to be nan
The output of the MultiBoxLoss is a single loss of all the batch [1], whereas the output of the LossNet is [32,1], so once these input are fed to the LossPredLoss it will return always return 0.0 as final loss module
The change that I've made regarding the SSD module is that I've attached the LossNet on this class and called the forward pass on the return of the SSD forward, like so:
returnoutput, self.loss_net(self.features), None
The None is for future work, for returning the embeddings
Hi, Thank you so much for your code.
I am unable to use the LossNet model for domain-specific datasets with different dimensions. It always says is forward()
Input dimension should be at least 3
or ndexError: Dimension out of range (expected to be in range of [-1, 0], but got -3)
@euphoria0-0 thanks for sharing the code base i have following query
the command "python main.py --task detection --dataset VOC0712
--subset None --num_epoch 300 --batch_size 32
--lr 0.001 --epoch_loss 240 --weights 1.0 --milestone " is incomplete with subset and milestone value not given can you confirm
Dataset voc0712 ? should have both voc2007 and voc2012 in our repo