Py-faster-rcnn: How to finetune pascal voc 2007 model with my images?

[limorbagizada]

How to finetune pascal voc 2007 model with my images?
I am new with deep learning and faster rcnn.
I have a folder of my images and I want to finetune it with the pascal voc 2007 model. How to do this?
Thanks

[deboc]

Hi,
There was already issues opened for fine-tuning the network on your own dataset.
If it can help I have made a tutorial on fine tuning py-faster-rcnn for a binary classifier.

But you should start with the caffe official website if you are totally new with caffe.

[ericromanenghi]

Close the issue please, there si a lot of issues in the repo about this.

[guysoft]

I suggest you close it _only_ if you link a duplicate issue. then you get all of the issues pointing to one.

[useebear]

@deboc
Hello Deboc, I followed your tutorial and did change all 84 to 8, but still came across such error during training as following. Thanks.

0923 16:21:49.104301 6578 sgd_solver.cpp:106] Iteration 9980, lr = 0.001
speed: 0.107s / iter
Process Process-3:
Traceback (most recent call last):
File "/usr/lib/python2.7/multiprocessing/process.py", line 258, in _bootstrap
self.run()
File "/usr/lib/python2.7/multiprocessing/process.py", line 114, in run
self._target(_self._args, *_self._kwargs)
File "./tools/train_faster_rcnn_alt_opt.py", line 195, in train_fast_rcnn
max_iters=max_iters)
File "/home/surview/Test/py-faster-rcnn/tools/../lib/fast_rcnn/train.py", line 160, in train_net
model_paths = sw.train_model(max_iters)
File "/home/surview/Test/py-faster-rcnn/tools/../lib/fast_rcnn/train.py", line 108, in train_model
model_paths.append(self.snapshot())
File "/home/surview/Test/py-faster-rcnn/tools/../lib/fast_rcnn/train.py", line 73, in snapshot
self.bbox_stds[:, np.newaxis])
ValueError: operands could not be broadcast together with shapes (84,1024) (8,1)

[deboc]

Hi,
I think it's a cache issue with your former batches.

[nassarofficial]

@useebear did @deboc reply solve your problem, because my cache is empty, and Im still getting this error!

[useebear]

@nassarofficial, thanks for the advice of @deboc. I have found the problem. It is because the prototxt is not in the correct folder. Check the folder.

[mohamed-ezz]

I had the same problem as @useebear , turned out I forgot to adjust the paths in the _solver.pt files. They have a path relative to the repo root dir so they point to models/pascal_voc/... by default.