Skip to content
Snippets Groups Projects
Commit 9353be88 authored by Andi Gerken's avatar Andi Gerken
Browse files

Updated models, datasets, index

parent e14bad89
No related branches found
No related tags found
No related merge requests found
python calculate_model_hashes.py
find /home/andi/blubber_workspace/model_server/public_html/ -type d -exec chmod 755 {} \;
find /home/andi/blubber_workspace/model_server/public_html/ -type f -exec chmod 655 {} \;
rsync -a --delete /home/andi/blubber_workspace/model_server/public_html/ login_zedat:public_html/model_server/
......@@ -90,23 +90,5 @@
"max_files": 20
},
"notes": "<span style='color:darkred;'>Should be regenerated to fit the current fish_models version</span>"
},
"pascal_vanilla_couzin":{
"path": "datasets/pascal_vanilla_couzin.pickle",
"date": "21.12.2021",
"dataset_config": {
"data_path": "pascal_vanilla_couzin",
"output_strings":["actions", "poses"]
},
"notes": "<span style='color:darkred;'>Should be regenerated to fit the current fish_models version</span>"
},
"pascal_180fov_couzin":{
"path": "datasets/pascal_180fov_couzin.pickle",
"date": "21.12.2021",
"dataset_config": {
"data_path": "pascal_180fov_couzin",
"output_strings":["actions", "poses"]
},
"notes": "<span style='color:darkred;'>Should be regenerated to fit the current fish_models version</span>"
}
}
}
\ No newline at end of file
......@@ -41,13 +41,13 @@
<th>Date</th>
<th>Dataset</th>
<th>Size</th>
<th>Generated Tracks</th>
<th>Generated Tracks</th>
<th>Evaluation</th>
<th>Video</th>
<th>Model</th>
</tr>
<?php
foreach($models_json as $model_name => $model_args){
foreach(array_reverse($models_json) as $model_name => $model_args){
echo "<tr>";
echo "<td>", $model_name, "</td>";
echo "<td>", $model_args["date"], "</td>";
......@@ -90,37 +90,89 @@
?>
</table>
<div>
<a href = "model_hashes.json">Model hashes (sha1)</a>
</div>
<div>
<h2>Models without training</h2>
Models which don't require training can be generated using the same <code>fish_models.ModelStorage.load_from_model_name("model.class.path(arg=42)")</code> function with the class path and arguments passed as a string (from version 0.3.7).
<ul>
<li><b>BasicFollowModel:</b> A model that always follows. The model turns can only operate with one partner fish in the tank. It will turn towards it smothly and has a constant speed. Normally distributed noise can be added to the turning angle.</br>
<a href='http://agerken.de/fish_models/models/basic_test_models.html#fish_models.models.basic_test_models.BasicFollowModel'><code>fish_models.models.basic_test_models.BasicFollowModel()</code></a>
<li><b>LiveFemaleFemaleConstantSpeedCouzinModel:</b> A constant speed couzin model with tuned parameters to fit the behavior of the live female female data. The behavior is non deterministic.</br>
Example tracks can be found in the documentation:</br>
<a href='http://agerken.de/fish_models/models/released/parametrized.html#fish_models.models.released.parametrized.LiveFemaleFemaleConstantSpeedCouzinModel'><code>fish_models.models.released.parametrized.LiveFemaleFemaleConstantSpeedCouzinModel()</code></a>
</ul>
<h1>Datasets</h1>
<table>
<tr>
<th>Name</th>
<th>Date</th>
<th>Data</th>
<th>Wallray</th>
<th>Fishbins</th>
<th>Output</th>
<th>Reduce dim</th>
<th>Max Files</th>
<th>Size</th>
<th>Download</th>
</tr>
<?php
<a href="model_hashes.json">Model hashes (sha1)</a>
</div>
<div>
<h2>Models without training</h2>
Models which don't require training can be generated using the same
<code>fish_models.ModelStorage.load_from_model_name("model.class.path(arg=42)")</code> function with the
class path and arguments passed as a string (from version 0.3.7).
<ul>
<li><b>BasicFollowModel:</b> A model that always follows. The model turns can only operate with one
partner fish in the tank. It will turn towards it smothly and has a constant speed. Normally
distributed noise can be added to the turning angle.</br>
<a
href='http://agerken.de/fish_models/models/basic_test_models.html#fish_models.models.basic_test_models.BasicFollowModel'><code>fish_models.models.basic_test_models.BasicFollowModel()</code></a>
<li><b>LiveFemaleFemaleConstantSpeedCouzinModel:</b> A constant speed couzin model with tuned parameters
to fit the behavior of the live female female data. The behavior is non deterministic.</br>
Example tracks can be found in the documentation:</br>
<a
href='http://agerken.de/fish_models/models/released/parametrized.html#fish_models.models.released.parametrized.LiveFemaleFemaleConstantSpeedCouzinModel'><code>fish_models.models.released.parametrized.LiveFemaleFemaleConstantSpeedCouzinModel()</code></a>
</ul>
<h1>Raw Data</h1>
<table>
<tr>
<th>Name</th>
<th>Author</th>
<th>Date</th>
<th>Tracking</th>
<th>Size</th>
<th>Folders [Files]</th>
<th>Evaluation</th>
<th>Download</th>
</tr>
<?php
foreach($raw_data_json as $raw_data_name => $rdarg){
if($_GET["raw_data"] == $raw_data_name){
echo "<tr id=$raw_data_name class=active>";
}else{
echo "<tr id=$raw_data_name>";
}
echo "<td>$raw_data_name</td>";
echo "<td>",$rdarg['author'],"</td>";
echo "<td>",$rdarg['date'],"</td>";
echo "<td>",$rdarg['tracking'],"</td>";
echo "<td>",round(filesize($rdarg['path']) / pow(1024, 2),1) ,"MB</td>";
echo "<td><code>";
$folder_str = "";
foreach($rdarg['folders'] as $folder){
$folder_str = $folder_str."$folder: [".(count(scandir($rdarg['path_unpacked']."/$folder")) - 2)."], ";
}
echo substr($folder_str, 0, -2);
echo "</code></td>";
if(file_exists("evaluation/raw_data/".$raw_data_name)){
echo "<td><a href='raw_data_details.php?raw_data=", $raw_data_name, "'>Show</a></td>";
}else{
echo"<td></td>";
}
echo "<td><a href='",$rdarg['path'],"'>Download</a></td>";
echo "</tr>";
if(array_key_exists("notes",$rdarg)){
echo "<tr><td></td><td colspan=9 class='textrow'>".$rdarg["notes"]."</td></tr>";
}
}
?>
</table>
<h1>Datasets (Deprecated?)</h1>
<table>
<tr>
<th>Name</th>
<th>Date</th>
<th>Data</th>
<th>Wallray</th>
<th>Fishbins</th>
<th>Output</th>
<th>Reduce dim</th>
<th>Max Files</th>
<th>Size</th>
<th>Download</th>
</tr>
<?php
foreach($datasets_json as $dataset_name => $darg){
if($_GET["dset"] == $dataset_name){
echo "<tr id=$dataset_name class=active>";
......@@ -154,53 +206,8 @@
}
}
?>
</table>
<h1>Raw Data</h1>
<table>
<tr>
<th>Name</th>
<th>Author</th>
<th>Date</th>
<th>Tracking</th>
<th>Size</th>
<th>Folders [Files]</th>
<th>Evaluation</th>
<th>Download</th>
</tr>
<?php
foreach($raw_data_json as $raw_data_name => $rdarg){
if($_GET["raw_data"] == $raw_data_name){
echo "<tr id=$raw_data_name class=active>";
}else{
echo "<tr id=$raw_data_name>";
}
echo "<td>$raw_data_name</td>";
echo "<td>",$rdarg['author'],"</td>";
echo "<td>",$rdarg['date'],"</td>";
echo "<td>",$rdarg['tracking'],"</td>";
echo "<td>",round(filesize($rdarg['path']) / pow(1024, 2),1) ,"MB</td>";
echo "<td><code>";
$folder_str = "";
foreach($rdarg['folders'] as $folder){
$folder_str = $folder_str."$folder: [".(count(scandir($rdarg['path_unpacked']."/$folder")) - 2)."], ";
}
echo substr($folder_str, 0, -2);
echo "</code></td>";
if(file_exists("evaluation/raw_data/".$raw_data_name)){
echo "<td><a href='raw_data_details.php?raw_data=", $raw_data_name, "'>Show</a></td>";
}else{
echo"<td></td>";
}
echo "<td><a href='",$rdarg['path'],"'>Download</a></td>";
echo "</tr>";
if(array_key_exists("notes",$rdarg)){
echo "<tr><td></td><td colspan=9 class='textrow'>".$rdarg["notes"]."</td></tr>";
}
}
?>
</table>
</div>
</table>
</div>
</body>
</html>
{"ReplayModel_v0.zip": "2e131cadbf30d78156a5ba4d0e2af0aeafdd6bc3", "GRU_with_conv_15fov.zip": "1be32ccdf9f28d428e11724d8e6f61e6fb1b16af", "KNNModel_v0.zip": "9ffa06b66a1519b7a44b325baddb5ed12e79ac33", "GRU_with_conv_51fov.zip": "f8f4d3729b4c94dfd8ff176802df7d7c1e142d4e", "GRU_without_conv_51_fov.zip": "c16a383de8dca96b5f4758fc66c70535985a0f15", "ClusterModel_v1_0.zip": "6148edf3820ba1a3227b192bd15a0f83d2c26d46", "ClassificationModel_v0.zip": "8945a0156dec3bc2a7269060c5b0a51d0525f0db", "ConvLinearDropoutClassifier_v0.zip": "8fc3561a93774f9187ae33ce9ed8ee226943533f", "ConvLinearDropoutClassifier_v1_trial_399.zip": "ec8fabecef782fc2e0f741dba7e0e2b9e711b2da", "ConvLinearDropoutClassifier_v1_trial_405.zip": "e283c494115e815ec8f5f064070bf0087654212b", "ClusterModel_v1_1.zip": "2de7e594c695b558644e19abd56d279bfb093fce", "ClusterModel_v0.zip": "52d184745f1f7879f7244845588376c314ef8b4a"}
\ No newline at end of file
{"ClusterModel_v0.zip": "52d184745f1f7879f7244845588376c314ef8b4a", "ClusterModel_v1_1.zip": "2de7e594c695b558644e19abd56d279bfb093fce", "ConvLinearDropoutClassifier_v1_trial_405.zip": "e283c494115e815ec8f5f064070bf0087654212b", "ConvLinearDropoutClassifier_v0.zip": "8fc3561a93774f9187ae33ce9ed8ee226943533f", "GRU_with_conv_51fov.zip": "f8f4d3729b4c94dfd8ff176802df7d7c1e142d4e", "ConvLinearDropoutClassifier_v1_trial_399.zip": "ec8fabecef782fc2e0f741dba7e0e2b9e711b2da", "ClassificationModel_v0.zip": "8945a0156dec3bc2a7269060c5b0a51d0525f0db", "GRU_with_conv_15fov.zip": "1be32ccdf9f28d428e11724d8e6f61e6fb1b16af", "ClusterModel_v1_0.zip": "6148edf3820ba1a3227b192bd15a0f83d2c26d46", "ConvLinearDropoutClassifier_v1_full_dataset.zip": "c95e411b80ed5d2f2aaa1faf865dde3f86a313db", "KNNModel_v0.zip": "9ffa06b66a1519b7a44b325baddb5ed12e79ac33", "GRU_without_conv_51_fov.zip": "c16a383de8dca96b5f4758fc66c70535985a0f15", "ReplayModel_v0.zip": "2e131cadbf30d78156a5ba4d0e2af0aeafdd6bc3"}
\ No newline at end of file
......@@ -98,5 +98,15 @@
"live_female_female/cherry_train",
"live_female_female/cherry_test"
]
},
"ConvLinearDropoutClassifier_v1_full_dataset": {
"path": "models/ConvLinearDropoutClassifier_v1_full_dataset.zip",
"date": "10.05.2022",
"notes": "The ConvLinearDropoutClsasifier trained to the full live_female_female dataset.",
"dataset": "live_female_female",
"evaluation_folders": [
"live_female_female/train",
"live_female_female/test"
]
}
}
\ No newline at end of file
}
0% Loading or .
You are about to add 0 people to the discussion. Proceed with caution.
Please register or to comment