neokt / car-damage-detective Goto Github PK
View Code? Open in Web Editor NEWAssessing car damage with convolution neural networks for a personal auto claims expedition use case
Assessing car damage with convolution neural networks for a personal auto claims expedition use case
I am trying out this code on my own data. However, my train_top_model() gives very good results with around 84% of accuracy. But while finetuning, my accuracy on validation doesn't change at all even after increasing the size of the data. I also tried reducing learning rate and keeping it very low but doesn't help. Any idea why this happens? Please help @neokt
Hi Neokt thank you for the great repo you have created. It is helping me a lot to understand the flow.
I am getting a little hard time to understand the flow of the directories you have created. Can you please explain the flow of it if possible. Thanks in advance.
Hi,
Unable to find bottleneck features.npy file along with code? any inputs what should it contain so we can code it manually?
Bharath
ValueError: Layer weight shape (3, 3, 3, 64) not compatible with provided weight shape (64, 3, 3, 3)
Just downloaded the dataset from google drive. However, it seem's the labels are not shipped w/ the bundle. The files are organised with subfolders. Do You have a .mat file for editing labels or something else ?
the link says that " the site can't be reached?"...please update the link!!...thanks for the wonderful presentation...i loved it...
Can you please provide a small step by step guide on getting this setup up and running. That would be really helpful @neokt .Thank you
What version of Python did you use in this project? If it's 2.7, I can't get Tensorflow to install using this version in Anaconda... Please help!
how to run?
unable to open file: name = 'static/models/d1_ft_model.h5', errno = 2, error message = 'No such file or directory'
Hi Neokt.
Greetings for the day. Your concept is really nice and I am trying to work on the same. I am very new to Machine Learning. Could you please help me on understanding this.
Hi, I am unable to access dataset from the given Google drive link. It get following error: Your connection is not private
Attackers might be trying to steal your information from goo.gl (for example, passwords, messages, or credit cards). Learn more
NET::ERR_CERT_AUTHORITY_INVALID
Please help me with the data access
Hello everyone while executing the code in jupyter notebook the following error was occured. please anybody can help.
NameError Traceback (most recent call last)
in ()
1 # do not rerun!!
----> 2 save_bottleneck_features(location)
in save_bottleneck_features(location)
2 datagen = ImageDataGenerator(rescale=1./255)
3
----> 4 model = load_vgg16()
5
6 generator = datagen.flow_from_directory(train_data_dir,
in load_vgg16(weights_path)
1 def load_vgg16(weights_path='../vgg16_weights.h5'):
----> 2 model = Sequential()
3 model.add(ZeroPadding2D((1,1),input_shape=(3, img_width, img_height)))
4 model.add(Convolution2D(64, 3, 3, activation='relu'))
5 model.add(ZeroPadding2D((1,1)))
NameError: global name 'Sequential' is not defined
can anybody help ton this issue please.
while calling the function in the python code as:
cat_counter = get_car_categories()
the called function definition is as follows:
def get_car_categories():
d = defaultdict(float)
img_list = os.listdir('data0')
for i, img_path in enumerate(img_list):
path = 'data0/'+img_path
img = prepare_image(path)
out = vgg16.predict(img)
top = get_predictions(out, top=5)
for j in top[0]:
d[j[0:2]] += j[2]
if i % 50 == 0:
print i, '/', len(img_list), 'complete'
return Counter(d)
IOError Traceback (most recent call last)
in ()
----> 1 cat_counter = get_car_categories()
in get_car_categories()
4 for i, img_path in enumerate(img_list):
5 path = 'data0/'+img_path
----> 6 img = prepare_image(path)
7 out = vgg16.predict(img)
8 top = get_predictions(out, top=5)
in prepare_image(img_path)
1 def prepare_image(img_path):
----> 2 img = load_img(img_path, target_size=(224, 224))
3 x = img_to_array(img)
4 x = np.expand_dims(x, axis=0)
5 x = preprocess_input(x)
/usr/local/lib/python2.7/dist-packages/keras/preprocessing/image.pyc in load_img(path, grayscale, target_size)
294 raise ImportError('Could not import PIL.Image. '
295 'The use of array_to_img
requires PIL.')
--> 296 img = pil_image.open(path)
297 if grayscale:
298 img = img.convert('L')
/usr/lib/python2.7/dist-packages/PIL/Image.pyc in open(fp, mode)
1994 if isPath(fp):
1995 filename = fp
-> 1996 fp = builtins.open(fp, "rb")
1997 else:
1998 filename = ""
IOError: [Errno 21] Is a directory: 'data0/.ipynb_checkpoints'
In this code from file 02-second_gate-damaged_or_whole :
f = h5py.File(weights_path)
for k in range(f.attrs['nb_layers']):
if k >= len(model.layers):
# we don't look at the last (fully-connected) layers in the savefile
break
g = f['layer_{}'.format(k)]
weights = [g['param_{}'.format(p)] for p in range(g.attrs['nb_params'])]
model.layers[k].set_weights(weights)
f.close()
print('VGG16 Model with partial weights loaded.')
else:
print('VGG16 Model with no weights Loaded.')
This error is occurring
Hello, where can I find one argument version of car_categories_gate ?
number2, bad_list2 = car_categories_gate(cat_list2)
And Where can I get these models?
second_gate = load_model('static/models/d1_ft_model.h5')
print( "Second gate loaded")
location_model = load_model('static/models/d2_ft_model.h5')
print( "Location model loaded")
severity_model = load_model('static/models/d3_ft_model.h5')
There are some bugs in the code, I was struggling to handle them.
A declarative, efficient, and flexible JavaScript library for building user interfaces.
๐ Vue.js is a progressive, incrementally-adoptable JavaScript framework for building UI on the web.
TypeScript is a superset of JavaScript that compiles to clean JavaScript output.
An Open Source Machine Learning Framework for Everyone
The Web framework for perfectionists with deadlines.
A PHP framework for web artisans
Bring data to life with SVG, Canvas and HTML. ๐๐๐
JavaScript (JS) is a lightweight interpreted programming language with first-class functions.
Some thing interesting about web. New door for the world.
A server is a program made to process requests and deliver data to clients.
Machine learning is a way of modeling and interpreting data that allows a piece of software to respond intelligently.
Some thing interesting about visualization, use data art
Some thing interesting about game, make everyone happy.
We are working to build community through open source technology. NB: members must have two-factor auth.
Open source projects and samples from Microsoft.
Google โค๏ธ Open Source for everyone.
Alibaba Open Source for everyone
Data-Driven Documents codes.
China tencent open source team.