Comments (3)
@zeehuilin
actually, this repo cloned from https://github.com/yhenon/keras-frcnn/.
so, I'm not implement first version of this code from scratch.
Actually, i also found this issue, so i was trying to implement validation process.
However, Faster R-CNN training process in this repo is starting from fire directory something like flow_from_generator()
for example, when training the Faster R-CNN using train_frcnn.py,
this script first parse Pascal VOC like annotation, (actually some another process also exist...) load image from parsed image path, and also extract ROIs(Bounding boxes) from annotation.
and then training Faster R-CNN using image, bbox information extracted above process.
this series of process is really time consuming.
for validate Object Detection Method, it need to calculate mean Average Precision.
you can explore mAP measuring process from measure_map.py
this process also time consuming.
So, Object Detection API such as Faster R-CNN implemented in https://github.com/tensorflow/models/tree/master/research/object_detection
is trained on TFRecord data type
To sum it up,
I was trying to implement validation process but it is actually difficult to implement with loading data from image file
from keras-fasterrcnn.
Thank you very much for your answer.
You've added a lot of BN (BatchNormalization) layers to xception.py.
In fact, I think there may be some problems with BN in Keras.I made a binary classification with your xception.py, but loss remained at 0.693 and did not converge. I also observed that in resnet50.py, the BN layer was FixedBatchNormalization.
In keras faster_rcnn (xception), RPN rpn_regr's loss declined slowly and detector_cls's loss declined quickly, which is quite different from keras faster_rcnn (vgg16) and keras faster_rcnn (res50).
Thank you again.
from keras-fasterrcnn.
i have question if someone could help
i am beginner ,and i training my data using Faster RCNN, but as mentioned the validation set code in 'train_frcnn.py' is not used, so don't know how to create validation set for training my data with this situation.
where can i add the validation code, and how?
from keras-fasterrcnn.
Related Issues (20)
- tensorflow.python.framework.errors_impl.InvalidArgumentError: slice index 14 of dimension 1 out of bounds. HOT 2
- Train Model Fail
- It doesn't train,Epoch 1/2000,and stop for one hour HOT 2
- How does val_loss works when train model
- How should I run the inference?
- Can anyone pls explain this region of code? Why are we only taking num_regions/2 positive and negative boxes?
- Pre-train for inception resnet v2 HOT 1
- Data Augmentation
- Plot the model
- Incompatible shapes
- custom dataset
- tf2 - from colab to local machine - problems with TimeDistributed
- stuck at epoch 1 for vgg HOT 1
- trained hdf5 file
- Unable to reproduce PASCAL VOC performance
- Exception: 'a' cannot be empty unless no samples are taken
- How to check Testing accuracy
- AttributeError: 'Node' object has no attribute 'output_masks' HOT 1
- ValueError: Attempt to convert a value (None) with an unsupported type (<class 'NoneType'>) to a Tensor. HOT 1
- Parallel Computing
Recommend Projects
-
React
A declarative, efficient, and flexible JavaScript library for building user interfaces.
-
Vue.js
🖖 Vue.js is a progressive, incrementally-adoptable JavaScript framework for building UI on the web.
-
Typescript
TypeScript is a superset of JavaScript that compiles to clean JavaScript output.
-
TensorFlow
An Open Source Machine Learning Framework for Everyone
-
Django
The Web framework for perfectionists with deadlines.
-
Laravel
A PHP framework for web artisans
-
D3
Bring data to life with SVG, Canvas and HTML. 📊📈🎉
-
Recommend Topics
-
javascript
JavaScript (JS) is a lightweight interpreted programming language with first-class functions.
-
web
Some thing interesting about web. New door for the world.
-
server
A server is a program made to process requests and deliver data to clients.
-
Machine learning
Machine learning is a way of modeling and interpreting data that allows a piece of software to respond intelligently.
-
Visualization
Some thing interesting about visualization, use data art
-
Game
Some thing interesting about game, make everyone happy.
Recommend Org
-
Facebook
We are working to build community through open source technology. NB: members must have two-factor auth.
-
Microsoft
Open source projects and samples from Microsoft.
-
Google
Google ❤️ Open Source for everyone.
-
Alibaba
Alibaba Open Source for everyone
-
D3
Data-Driven Documents codes.
-
Tencent
China tencent open source team.
from keras-fasterrcnn.