Comments (16)
Looks like it's fine.
Mind you have quite high loss connected with classification to a class and low loss value in finding proposal. Looks like it's not a big deal in finding proposals, but rather with setting a correct class for a proposal.
Maybe try first with lower number of classes? Like give only 5 classes in the beginning to check if it'll work at all. Moreover, your photos looks like to be huge (especially long) - try to resize them down? I'm using 768x768 for example
from keras-frcnn.
Has the problem got solved? Because I'm facing the same issue.
from keras-frcnn.
I assume you trained model using train_frcnn.py, but did not run test_frcnn.py. The first file trained model using your images and created weights file, but you need to apply run second file, as it'll look for objects in the images.
For me it's working fine
from keras-frcnn.
I first run train_frcnn.py after i got the weight file , then i run test_frcnn to object detection. did i miss any step here?
from keras-frcnn.
No, maybe it's not enough training time. What is your best current loose for now?
from keras-frcnn.
from keras-frcnn.
actually i stopped in 1500 epochs ran , i think that will be enough
from keras-frcnn.
Do you set any bound box as background (bg)? I assume you're passing the entry data as txt file, right?
from keras-frcnn.
ya i use annotation as txt file
from keras-frcnn.
Do you set any bound box as background (bg)? Having entries in txt like
abcd.jpg,0,10,20,30,bg
from keras-frcnn.
This is my sample annotate.txt file
NumberPlate/JPEGImages/SKP3858H.jpg,159,193,450,567,S
from keras-frcnn.
is there any wrong? can you give any advice
from keras-frcnn.
i got some character detection after 3000 epcohs, but still it is not recognized new unknown images.
I have worry about this algorithm implementation.
from keras-frcnn.
i don`t have GPU in my laptop is it any issue in this process?
from keras-frcnn.
are you using the correct coordinate system? a point in matrix image[y,x,3] should be refer as x,y in annotate.txt.
from keras-frcnn.
Has the problem got solved? Because I'm facing the same issue.
I'm doing for you
from keras-frcnn.
Related Issues (20)
- Error with get_data(input_path) function of simple_parser.py HOT 3
- Wrong batch normalization axis for classifier layers, using tf
- Please fix the code to print the number of training samples in line 112
- ValueError: Attempt to convert a value (None) with an unsupported type (<class 'NoneType'>) to a Tensor. HOT 4
- input image must be of non-zero size
- AttributeError: 'NoneType' object has no attribute 'shape' HOT 3
- Error if there are a lot of boxes, i guess
- No detection result show
- Meaning of epoch_length HOT 1
- Where has the function `img_length_calc_function` used in the `calc_rpn` function been defined?
- Parallel Computing
- Working with custom model
- usage of measure_map HOT 1
- SyntaxError HOT 2
- Input 0 of layer "model" is incompatible with the layer HOT 3
- Problem in Faster R-CNN Algorithm for Object Detection
- Update your frcnn please
- Could not create cudnn handle: CUDNN_STATUS_ALLOC_FAILED
- How to convert trained .hdf5 model to a single TensorFlow .pb file HOT 2
- Exception: index 1000 is out of bounds for axis 0 with size 1000 HOT 11
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-frcnn.