Comments (9)
pierluigiferrari
commented on May 23, 2024
I haven't seen this issue before and can't think of a plausible explanation yet. It's strange that the size of the box is off but the center of the box is still roughly in the right place. I'll spend some more time on this. A few questions:
- Is this the result after using one of the Jupyter notebooks more or less out of the box or did you change a lot of code?
- Is this behavior consistent, i.e. do you see the same phenomenon on multiple images?
from ssd_keras.
lababidi
commented on May 23, 2024
I went ahead and retained and couldn't reproduce it. I must have tweaked a
parameter or one of your layers. I was attempting to get save_model and
load_model writing by tweaking your custom layers. I almost have that
working. Also I found a way to load the VGG weights before training. Are
you interested in that?
…On Mon, Sep 18, 2017 at 9:08 PM Pierluigi Ferrari ***@***.***> wrote:
I haven't seen this issue before and can't think of a plausible
explanation yet. It's strange that the size of the box is off but the
center of the box is still roughly in the right place. I'll spend some more
time on this. A few questions:
1. Is this the result after using one of the training notebooks more
or less out of the box or did you change a lot of code?
2. Is this behavior consistent, i.e. do you see the same phenomenon on
multiple images?
—
You are receiving this because you authored the thread.
Reply to this email directly, view it on GitHub
<#7 (comment)>,
or mute the thread
<https://github.com/notifications/unsubscribe-auth/AAePJyZOmRrRTbtUk2piJ1gn4S77-Sgsks5sjxQPgaJpZM4Pbqv_>
.
from ssd_keras.
pierluigiferrari
commented on May 23, 2024
Ah, I'm glad you couldn't reproduce it. And yes, if you have a h5 version of the pretrained VGG weights used in the original Caffe implementation that are compatible with this implementation, that would be great :)
from ssd_keras.
lababidi
commented on May 23, 2024
it's approximately 90MB
from ssd_keras.
lababidi
commented on May 23, 2024
just model.load_weights('vgg.h5) should work. let me know if it doesn't.
from ssd_keras.
pierluigiferrari
commented on May 23, 2024
I managed to load the VGG weights into the SSD300 model, but I had to rename the predictor layers in the model, because otherwise it tried to load weights into the predictor layers but complained that the shapes wouldn't match. It's weird that there seem to be matching layer names for the SSD300 predictor layers in the weights file.
Do you know the origin of these weights? Also, were they trained on classification or localization / object detection? Most trained weights for VGG out there were trained on classification tasks.
from ssd_keras.
lababidi
commented on May 23, 2024
I took the VGG weights from keras and exported them as numpy arrays. I
modified your ssd_300 function to load these numpy arrays from file per
layer. I also set the first 3 conv groups to untrainable. I exported this
model's weights and it's what you have.
I'm happy to provide any other help if you need.
…On Wed, Sep 27, 2017 at 9:35 AM Pierluigi Ferrari ***@***.***> wrote:
I managed to load the VGG weights into the SSD300 model, but I had to
rename the predictor layers in the model, because otherwise it tried to
load weights into the predictor layers but complained that the shapes
wouldn't match. It's weird that there seem to be matching layer names for
the SSD300 predictor layers in the weights file.
Do you know the origin of these weights? Also, were they trained on
classification or localization / object detection? Most trained weights for
VGG out there were trained on classification tasks.
—
You are receiving this because you authored the thread.
Reply to this email directly, view it on GitHub
<#7 (comment)>,
or mute the thread
<https://github.com/notifications/unsubscribe-auth/AAePJ603ca9ee4y5ACX7pXYrT_cHXDHHks5smk8mgaJpZM4Pbqv_>
.
from ssd_keras.
wushilian
commented on May 23, 2024
Hi,I use keras Apllication model,and it work well.after download keras offered model,you just need model.load_weights('vgg.h5',True)
from ssd_keras.
pierluigiferrari
commented on May 23, 2024
Thanks Mahmoud! I ported my own VGG-16 weights since yours don't include the convolutionalized fc6 and fc7 weights. I followed the kernel-subsampling procedure of the original implementation for the fc6 and fc7 weights. I trained SSD300 on Pascal VOC with these weights and it works very well. Download link is in the README if you're interested
from ssd_keras.
Related Issues (20)
- InvalidArgumentError when compiling model with ssd_loss HOT 1
- WARNING:tensorflow:Gradients do not exist for variables ['conv4_3/bias:0',...] when minimizing the loss. HOT 1
- "Invalid argument: Index out of range using input dim 0; input has only 0 dims" during ssd300 model training
- load weight
- ValueError: Error when checking input: expected input_3 to have 4 dimensions, but got array with shape
- While training I got training terminate error . Epoch 00001: LearningRateScheduler setting learning rate to 0.001. 1/10 [==>...........................] - ETA: 4:08 - loss: nanBatch 0: Invalid loss, terminating training Epoch 00001: saving model to ssd512_URPC2018_epoch-01.h5 Process finished with exit code 0
- ValueError: An operation has `None` for gradient. Please make sure that all of your ops have a gradient defined (i.e. are differentiable). Common ops without gradient: K.argmax, K.round, K.eval.
- ValueError: Layer model expects 1 input(s), but it received 2 input tensors. Inputs received: [<tf.Tensor 'IteratorGetNext:0' shape=(None, None, None, None) dtype=uint8>, <tf.Tensor 'IteratorGetNext:1' shape=(None, None, None) dtype=float32>] HOT 23
- Parameters of the model HOT 1
- Bouding boxes predictions are concentrated in left top corner HOT 1
- Ambiguous dimension while trying to load weights.
- Urgent!! Invalid Loss HOT 4
- What are the requirements to run this code?. HOT 1
- Pascal VOC Training Person Detection
- The device being used is CPU while capturing image from webcam. How do I use my GPU for processing instead?
- Label error during Coco Training HOT 1
- TypeError: Expected any non-tensor type, got a tensor instead.
- Changes make the code work in 2023 HOT 2
- custom SSD300 model
- error while training with custom dataset in COCO format
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 ssd_keras.